{
"version": "1.0",
"updatedAt": "2026-09-12T08:11:58Z",
"totalProviders": 213,
"totalModels": 7751,
"providers": {
"subconscious": {
"id": "subconscious",
"name": "Subconscious",
"baseURL": "https://api.subconscious.dev/v1",
"npm": "@ai-sdk/anthropic",
"swiftDriver": "anthropicMessages",
"env": [
"SUBCONSCIOUS_API_KEY"
],
"doc": "https://docs.subconscious.dev",
"modelCount": 2,
"models": {
"subconscious/glm-5.2": {
"id": "subconscious/glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"subconscious/subconscious/glm-5.2\", apiKey: processEnvironment[\"SUBCONSCIOUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.subconscious.dev/v1\")!,\n apiKey: processEnvironment[\"SUBCONSCIOUS_API_KEY\"]\n)\nlet session = provider.model(\"subconscious/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"subconscious/tim-qwen3.6-27b": {
"id": "subconscious/tim-qwen3.6-27b",
"name": "TIM-Qwen3.6 27B",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-11",
"last_updated": "2026-05-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"input": 8192,
"output": 5000
},
"cost": {
"input": 0.3,
"output": 3,
"cache_read": 0.15
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"subconscious/subconscious/tim-qwen3.6-27b\", apiKey: processEnvironment[\"SUBCONSCIOUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.subconscious.dev/v1\")!,\n apiKey: processEnvironment[\"SUBCONSCIOUS_API_KEY\"]\n)\nlet session = provider.model(\"subconscious/tim-qwen3.6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"tokengo": {
"id": "tokengo",
"name": "TokenGo",
"baseURL": "https://api.tokengo.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"TOKENGO_API_KEY"
],
"doc": "https://www.tokengo.com/docs",
"modelCount": 13,
"models": {
"qwen/qwen3.5-397b-a17b": {
"id": "qwen/qwen3.5-397b-a17b",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 2.65,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/qwen/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5": {
"id": "minimax/minimax-m2.5",
"name": "MiniMax-M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/minimax/minimax-m2.5\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash": {
"id": "deepseek/deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.098,
"output": 0.196,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/deepseek/deepseek-v4-flash\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.2": {
"id": "deepseek/deepseek-v3.2",
"name": "DeepSeek V3.2",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 64000
},
"cost": {
"input": 0.2174,
"output": 0.326,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/deepseek/deepseek-v3.2\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro": {
"id": "deepseek/deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.435,
"output": 0.87
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/deepseek/deepseek-v4-pro\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.1": {
"id": "deepseek/deepseek-v3.1",
"name": "DeepSeek-V3.1",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.19,
"output": 0.71,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/deepseek/deepseek-v3.1\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.6": {
"id": "moonshotai/kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/moonshotai/kimi-k2.6\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k3": {
"id": "moonshotai/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/moonshotai/kimi-k3\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.2": {
"id": "z-ai/glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/z-ai/glm-5.2\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.3-flash": {
"id": "z-ai/glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.075,
"output": 0.025,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/z-ai/glm-5.3-flash\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5": {
"id": "z-ai/glm-5",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.89,
"output": 3.2647,
"cache_read": 0.2226
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/z-ai/glm-5\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.1": {
"id": "z-ai/glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/z-ai/glm-5.1\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.3": {
"id": "z-ai/glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokengo/z-ai/glm-5.3\", apiKey: processEnvironment[\"TOKENGO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokengo.com/v1\")!,\n apiKey: processEnvironment[\"TOKENGO_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"modelis": {
"id": "modelis",
"name": "Modelis",
"baseURL": "https://modelishub.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"MODELIS_API_KEY"
],
"doc": "https://modelishub.com/pricing",
"modelCount": 9,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelis/claude-sonnet-4-6\", apiKey: processEnvironment[\"MODELIS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://modelishub.com/v1\")!,\n apiKey: processEnvironment[\"MODELIS_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.0983,
"output": 0.1966
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelis/deepseek-v4-flash\", apiKey: processEnvironment[\"MODELIS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://modelishub.com/v1\")!,\n apiKey: processEnvironment[\"MODELIS_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelis/claude-fable-5\", apiKey: processEnvironment[\"MODELIS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://modelishub.com/v1\")!,\n apiKey: processEnvironment[\"MODELIS_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
},
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelis/claude-opus-4-8\", apiKey: processEnvironment[\"MODELIS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://modelishub.com/v1\")!,\n apiKey: processEnvironment[\"MODELIS_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.435,
"output": 0.87
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelis/deepseek-v4-pro\", apiKey: processEnvironment[\"MODELIS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://modelishub.com/v1\")!,\n apiKey: processEnvironment[\"MODELIS_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelis/gemini-2.5-pro\", apiKey: processEnvironment[\"MODELIS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://modelishub.com/v1\")!,\n apiKey: processEnvironment[\"MODELIS_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"max"
]
},
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelis/gemini-2.5-flash\", apiKey: processEnvironment[\"MODELIS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://modelishub.com/v1\")!,\n apiKey: processEnvironment[\"MODELIS_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.7-max": {
"id": "qwen/qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 3,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelis/qwen/qwen3.7-max\", apiKey: processEnvironment[\"MODELIS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://modelishub.com/v1\")!,\n apiKey: processEnvironment[\"MODELIS_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.7-plus": {
"id": "qwen/qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.768,
"output": 3.072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelis/qwen/qwen3.7-plus\", apiKey: processEnvironment[\"MODELIS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://modelishub.com/v1\")!,\n apiKey: processEnvironment[\"MODELIS_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"bothub": {
"id": "bothub",
"name": "Bothub",
"baseURL": "https://openai.bothub.ru/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"BOTHUB_API_KEY"
],
"doc": "https://bothub.ru/models",
"modelCount": 8,
"models": {
"deepseek-v4-pro-0813": {
"id": "deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 1.61,
"output": 4.84
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"bothub/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"BOTHUB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://openai.bothub.ru/v1\")!,\n apiKey: processEnvironment[\"BOTHUB_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.1,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"bothub/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"BOTHUB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://openai.bothub.ru/v1\")!,\n apiKey: processEnvironment[\"BOTHUB_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-luna": {
"id": "gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.06,
"output": 0.37
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"bothub/gpt-5.6-luna\", apiKey: processEnvironment[\"BOTHUB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://openai.bothub.ru/v1\")!,\n apiKey: processEnvironment[\"BOTHUB_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.12,
"output": 0.44
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"bothub/glm-5.3-flash\", apiKey: processEnvironment[\"BOTHUB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://openai.bothub.ru/v1\")!,\n apiKey: processEnvironment[\"BOTHUB_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1.3-contributor": {
"id": "muse-spark-1.3-contributor",
"name": "Muse Spark 1.3 Contributor",
"description": "Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It improves long-horizon agent collaboration, instruction following, and coding efficiency relative to Muse Spark 1.2.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 943718
},
"cost": {
"input": 0.1,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"bothub/muse-spark-1.3-contributor\", apiKey: processEnvironment[\"BOTHUB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://openai.bothub.ru/v1\")!,\n apiKey: processEnvironment[\"BOTHUB_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1.3-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-ultra-550b-a55b:free": {
"id": "nemotron-3-ultra-550b-a55b:free",
"name": "Nemotron 3 Ultra (free)",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"bothub/nemotron-3-ultra-550b-a55b:free\", apiKey: processEnvironment[\"BOTHUB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://openai.bothub.ru/v1\")!,\n apiKey: processEnvironment[\"BOTHUB_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-ultra-550b-a55b:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it:free": {
"id": "gemma-4-31b-it:free",
"name": "Gemma 4 31B IT (free)",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"bothub/gemma-4-31b-it:free\", apiKey: processEnvironment[\"BOTHUB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://openai.bothub.ru/v1\")!,\n apiKey: processEnvironment[\"BOTHUB_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.72,
"output": 5.41
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"bothub/glm-5.3\", apiKey: processEnvironment[\"BOTHUB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://openai.bothub.ru/v1\")!,\n apiKey: processEnvironment[\"BOTHUB_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"greenpt": {
"id": "greenpt",
"name": "GreenPT",
"baseURL": "https://api.greenpt.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"GREENPT_API_KEY"
],
"doc": "https://docs.greenpt.ai",
"modelCount": 40,
"models": {
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.1596,
"output": 0.399,
"cache_read": 0.0456
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-caveman-ultra": {
"id": "glm-5.2-caveman-ultra",
"name": "GLM-5.2 Caveman Ultra",
"description": "glm-5.2 carrying a built-in ruleset that compresses prose, keeping code and technical detail verbatim. The most aggressive tier, close to answer-only. Same upstream model and price per token as glm-5.2, with fewer output tokens.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.254,
"output": 5.016,
"cache_read": 0.3135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.2-caveman-ultra\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-caveman-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-ponytail-ultra": {
"id": "glm-5.2-ponytail-ultra",
"name": "GLM-5.2 Ponytail Ultra",
"description": "glm-5.2 carrying a built-in ruleset that compresses generated code, preferring platform features over custom code. The most aggressive tier, close to answer-only. Same upstream model and price per token as glm-5.2, with fewer output tokens.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.254,
"output": 5.016,
"cache_read": 0.3135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.2-ponytail-ultra\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-ponytail-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-honey-ultra": {
"id": "glm-5.2-honey-ultra",
"name": "GLM-5.2 Honey Ultra",
"description": "glm-5.2 carrying a built-in ruleset that compresses both generated code and prose. The most aggressive tier, close to answer-only. Same upstream model and price per token as glm-5.2, with fewer output tokens.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.254,
"output": 5.016,
"cache_read": 0.3135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.2-honey-ultra\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-honey-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.7524,
"output": 4.275,
"cache_read": 0.2508
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/kimi-k2.6\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.254,
"output": 5.016,
"cache_read": 0.3135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.2\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.5": {
"id": "minimax-m2.5",
"name": "MiniMax-M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.1938,
"output": 1.129,
"cache_read": 0.0627
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/minimax-m2.5\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code": {
"id": "kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.9006,
"output": 4.389,
"cache_read": 0.1881
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/kimi-k2.7-code\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"green-l": {
"id": "green-l",
"name": "Green L",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-06-20",
"last_updated": "2025-06-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.285,
"output": 0.912
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/green-l\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"green-l\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"devstral-2-123b-instruct-2512": {
"id": "devstral-2-123b-instruct-2512",
"name": "Devstral 2",
"description": "Mistral's coding-agent model for repository work, terminal tasks, and software fixes",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2025-12-09",
"last_updated": "2025-12-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 16384
},
"cost": {
"input": 0.57,
"output": 2.736
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/devstral-2-123b-instruct-2512\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"devstral-2-123b-instruct-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4.1-flash": {
"id": "deepseek-v4.1-flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.255552,
"output": 1.27776,
"cache_read": 0.0127776
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/deepseek-v4.1-flash\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4.1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-honey-lite": {
"id": "glm-5.2-honey-lite",
"name": "GLM-5.2 Honey Lite",
"description": "glm-5.2 carrying a built-in ruleset that compresses both generated code and prose. The gentlest tier: it cuts filler only and keeps the explanation intact. Same upstream model and price per token as glm-5.2, with fewer output tokens.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.254,
"output": 5.016,
"cache_read": 0.3135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.2-honey-lite\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-honey-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-30b-a3b-instruct": {
"id": "qwen3-coder-30b-a3b-instruct",
"name": "Qwen3-Coder 30B-A3B Instruct",
"description": "Smaller Qwen coder for efficient local agents and repo-level fixes",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.285,
"output": 1.083
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"green-l-raw": {
"id": "green-l-raw",
"name": "Green L Raw",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-06-20",
"last_updated": "2025-06-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.285,
"output": 0.912
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/green-l-raw\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"green-l-raw\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-397b-a17b": {
"id": "qwen3.5-397b-a17b",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.798,
"output": 4.959
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-caveman": {
"id": "glm-5.2-caveman",
"name": "GLM-5.2 Caveman",
"description": "glm-5.2 carrying a built-in ruleset that compresses prose, keeping code and technical detail verbatim. The middle tier, and the ruleset as its authors wrote it. Same upstream model and price per token as glm-5.2, with fewer output tokens.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.254,
"output": 5.016,
"cache_read": 0.3135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.2-caveman\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-caveman\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3.762,
"output": 18.81,
"cache_read": 0.9405
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/kimi-k3\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-3-27b-it": {
"id": "gemma-3-27b-it",
"name": "Gemma 3 27B",
"description": "Google Gemma 3 multimodal model for chat, reasoning, and image understanding",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-03-12",
"last_updated": "2025-03-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40000,
"output": 8192
},
"cost": {
"input": 0.342,
"output": 0.684
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/gemma-3-27b-it\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-3-27b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-35b-a3b": {
"id": "qwen3.6-35b-a3b",
"name": "Qwen3.6 35B-A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.342,
"output": 2.052
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/qwen3.6-35b-a3b\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"green-s": {
"id": "green-s",
"name": "Green S",
"description": "GreenPT speech-to-text model for pre-recorded and live transcription",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-01",
"last_updated": "2025-01",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 8192
},
"cost": {
"input": 0.00437,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/green-s\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"green-s\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.127754,
"output": 0.511016,
"cache_read": 0.0255508
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.3-flash\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-caveman-lite": {
"id": "glm-5.2-caveman-lite",
"name": "GLM-5.2 Caveman Lite",
"description": "glm-5.2 carrying a built-in ruleset that compresses prose, keeping code and technical detail verbatim. The gentlest tier: it cuts filler only and keeps the explanation intact. Same upstream model and price per token as glm-5.2, with fewer output tokens.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.254,
"output": 5.016,
"cache_read": 0.3135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.2-caveman-lite\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-caveman-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-medium-3.5-128b": {
"id": "mistral-medium-3.5-128b",
"name": "Mistral Medium 3.5",
"description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
"family": "mistral-medium",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-29",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 2.052,
"output": 10.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/mistral-medium-3.5-128b\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"mistral-medium-3.5-128b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-ponytail": {
"id": "glm-5.2-ponytail",
"name": "GLM-5.2 Ponytail",
"description": "glm-5.2 carrying a built-in ruleset that compresses generated code, preferring platform features over custom code. The middle tier, and the ruleset as its authors wrote it. Same upstream model and price per token as glm-5.2, with fewer output tokens.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.254,
"output": 5.016,
"cache_read": 0.3135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.2-ponytail\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-ponytail\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma4": {
"id": "gemma4",
"name": "gemma4",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.57,
"output": 1.71
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/gemma4\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-3.2-24b-instruct-2506": {
"id": "mistral-small-3.2-24b-instruct-2506",
"name": "Mistral Small 3.2",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-06-20",
"last_updated": "2025-06-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.228,
"output": 0.456
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/mistral-small-3.2-24b-instruct-2506\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small-3.2-24b-instruct-2506\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-honey": {
"id": "glm-5.2-honey",
"name": "GLM-5.2 Honey",
"description": "glm-5.2 carrying a built-in ruleset that compresses both generated code and prose. The middle tier, and the ruleset as its authors wrote it. Same upstream model and price per token as glm-5.2, with fewer output tokens.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.254,
"output": 5.016,
"cache_read": 0.3135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.2-honey\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-honey\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"status": "deprecated",
"cost": {
"input": 1.756,
"output": 5.518
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.1\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-ponytail-lite": {
"id": "glm-5.2-ponytail-lite",
"name": "GLM-5.2 Ponytail Lite",
"description": "glm-5.2 carrying a built-in ruleset that compresses generated code, preferring platform features over custom code. The gentlest tier: it cuts filler only and keeps the explanation intact. Same upstream model and price per token as glm-5.2, with fewer output tokens.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.254,
"output": 5.016,
"cache_read": 0.3135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.2-ponytail-lite\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-ponytail-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-instruct-2507": {
"id": "qwen3-235b-a22b-instruct-2507",
"name": "Qwen3 235B A22B Instruct 2507",
"description": "Qwen3 235B MoE instruct model for long-context multilingual chat and reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-21",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 1.026,
"output": 3.078
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/qwen3-235b-a22b-instruct-2507\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"green-r-raw": {
"id": "green-r-raw",
"name": "Green R Raw",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.399,
"output": 1.083
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/green-r-raw\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"green-r-raw\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.228,
"output": 0.798
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/gpt-oss-120b\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"holo2-30b-a3b": {
"id": "holo2-30b-a3b",
"name": "Holo2 30B A3B",
"description": "H Company Holo2 vision model for GUI navigation and computer-use agents",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-11",
"last_updated": "2025-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 22016,
"output": 16384
},
"cost": {
"input": 0.399,
"output": 0.969
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/holo2-30b-a3b\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"holo2-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voxtral-small-24b-2507": {
"id": "voxtral-small-24b-2507",
"name": "Voxtral Small 24B",
"description": "Mistral Voxtral audio-understanding model for speech and transcription tasks",
"family": "mistral",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-07-15",
"last_updated": "2025-07-15",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0.228,
"output": 0.513
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/voxtral-small-24b-2507\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"voxtral-small-24b-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.27754,
"output": 5.11016,
"cache_read": 0.319385
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/glm-5.3\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"green-s-pro": {
"id": "green-s-pro",
"name": "Green S Pro",
"description": "GreenPT advanced speech-to-text model with multilingual transcription support",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-02",
"last_updated": "2025-02",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 8192
},
"cost": {
"input": 0.00437,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/green-s-pro\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"green-s-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6-fast": {
"id": "kimi-k2.6-fast",
"name": "Kimi K2.6 Fast",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"status": "deprecated",
"cost": {
"input": 1.655,
"output": 8.778
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/kimi-k2.6-fast\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"pixtral-12b-2409": {
"id": "pixtral-12b-2409",
"name": "Pixtral 12B",
"description": "Mistral vision-language model for image understanding and multimodal chat",
"family": "pixtral",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2024-09-01",
"last_updated": "2024-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.285,
"output": 0.285
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/pixtral-12b-2409\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"pixtral-12b-2409\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"green-r": {
"id": "green-r",
"name": "Green R",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.399,
"output": 1.083
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/green-r\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"green-r\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-70b-instruct": {
"id": "llama-3.3-70b-instruct",
"name": "Llama-3.3-70B-Instruct",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 100000,
"output": 16384
},
"cost": {
"input": 1.254,
"output": 1.254
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"greenpt/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"GREENPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.greenpt.ai/v1\")!,\n apiKey: processEnvironment[\"GREENPT_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"qiniu-ai": {
"id": "qiniu-ai",
"name": "Qiniu",
"baseURL": "https://api.qnaigc.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"QINIU_API_KEY"
],
"doc": "https://developer.qiniu.com/aitokenapi",
"modelCount": 91,
"models": {
"glm-4.5-air": {
"id": "glm-4.5-air",
"name": "GLM 4.5 Air",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131000,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/glm-4.5-air\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5-air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kling-v2-6": {
"id": "kling-v2-6",
"name": "Kling-V2 6",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-01-13",
"last_updated": "2026-01-13",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 99999999,
"output": 99999999
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/kling-v2-6\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"kling-v2-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-next-80b-a3b-thinking": {
"id": "qwen3-next-80b-a3b-thinking",
"name": "Qwen3 Next 80B A3B Thinking",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-12",
"last_updated": "2025-09-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-next-80b-a3b-thinking\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-next-80b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3": {
"id": "deepseek-v3",
"name": "DeepSeek-V3",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-13",
"last_updated": "2025-08-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/deepseek-v3\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-thinking-2507": {
"id": "qwen3-235b-a22b-thinking-2507",
"name": "Qwen3 235B A22B Thinking 2507",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-12",
"last_updated": "2025-08-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-235b-a22b-thinking-2507\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.0-pro-image-preview": {
"id": "gemini-3.0-pro-image-preview",
"name": "Gemini 3.0 Pro Image Preview",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-11-20",
"last_updated": "2025-11-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 8192
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/gemini-3.0-pro-image-preview\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.0-pro-image-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-image": {
"id": "gemini-2.5-flash-image",
"name": "Gemini 2.5 Flash Image",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-10-22",
"last_updated": "2025-10-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 8192
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/gemini-2.5-flash-image\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-next-80b-a3b-instruct": {
"id": "qwen3-next-80b-a3b-instruct",
"name": "Qwen3 Next 80B A3B Instruct",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-12",
"last_updated": "2025-09-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.0-flash": {
"id": "gemini-2.0-flash",
"name": "Gemini 2.0 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 8192
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/gemini-2.0-flash\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.0-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-3.5-sonnet": {
"id": "claude-3.5-sonnet",
"name": "Claude 3.5 Sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-09",
"last_updated": "2025-09-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8200
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/claude-3.5-sonnet\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"claude-3.5-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2.0-mini": {
"id": "doubao-seed-2.0-mini",
"name": "Doubao Seed 2.0 Mini",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/doubao-seed-2.0-mini\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2.0-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-turbo": {
"id": "qwen-turbo",
"name": "Qwen-Turbo",
"description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen-turbo\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite": {
"id": "gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 64000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.0-pro-preview": {
"id": "gemini-3.0-pro-preview",
"name": "Gemini 3.0 Pro Preview",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/gemini-3.0-pro-preview\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.0-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-1.6": {
"id": "doubao-seed-1.6",
"name": "Doubao-Seed 1.6",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-15",
"last_updated": "2025-08-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/doubao-seed-1.6\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-1.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-32b": {
"id": "qwen3-32b",
"name": "Qwen3 32B",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 40000,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-32b\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-flash": {
"id": "mimo-v2-flash",
"name": "Mimo-V2-Flash",
"description": "MiMo flash model for fast multimodal assistance and agent workflows",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12-01",
"release_date": "2025-12-16",
"last_updated": "2026-02-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/mimo-v2-flash\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.0-flash-lite": {
"id": "gemini-2.0-flash-lite",
"name": "Gemini 2.0 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 8192
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/gemini-2.0-flash-lite\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.0-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-4.5-opus": {
"id": "claude-4.5-opus",
"name": "Claude 4.5 Opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-11-25",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 200000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/claude-4.5-opus\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"claude-4.5-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-20b": {
"id": "gpt-oss-20b",
"name": "gpt-oss-20b",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-06",
"last_updated": "2025-08-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/gpt-oss-20b\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2.5-vl-72b-instruct": {
"id": "qwen2.5-vl-72b-instruct",
"name": "Qwen 2.5 VL 72B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen2.5-vl-72b-instruct\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen2.5-vl-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-4.0-opus": {
"id": "claude-4.0-opus",
"name": "Claude 4.0 Opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/claude-4.0-opus\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"claude-4.0-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-max-preview": {
"id": "qwen3-max-preview",
"name": "Qwen3 Max Preview",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-06",
"last_updated": "2025-09-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 64000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-max-preview\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-1.5-pro-32k": {
"id": "doubao-1.5-pro-32k",
"name": "Doubao 1.5 Pro 32k",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 12000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/doubao-1.5-pro-32k\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"doubao-1.5-pro-32k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2.0-code": {
"id": "doubao-seed-2.0-code",
"name": "Doubao Seed 2.0 Code",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/doubao-seed-2.0-code\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2.0-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-397b-a17b": {
"id": "qwen3.5-397b-a17b",
"name": "Qwen3.5 397B A17B",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-22",
"last_updated": "2026-02-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 64000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1": {
"id": "deepseek-r1",
"name": "DeepSeek-R1",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/deepseek-r1\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2.5-vl-7b-instruct": {
"id": "qwen2.5-vl-7b-instruct",
"name": "Qwen 2.5 VL 7B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen2.5-vl-7b-instruct\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen2.5-vl-7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-4.1-opus": {
"id": "claude-4.1-opus",
"name": "Claude 4.1 Opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-06",
"last_updated": "2025-08-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/claude-4.1-opus\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"claude-4.1-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-30b-a3b-thinking-2507": {
"id": "qwen3-30b-a3b-thinking-2507",
"name": "Qwen3 30b A3b Thinking 2507",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-04",
"last_updated": "2026-02-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 126000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-30b-a3b-thinking-2507\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-30b-a3b-thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1-0528": {
"id": "deepseek-r1-0528",
"name": "DeepSeek-R1-0528",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/deepseek-r1-0528\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1-0528\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-vl-max-2025-01-25": {
"id": "qwen-vl-max-2025-01-25",
"name": "Qwen VL-MAX-2025-01-25",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen-vl-max-2025-01-25\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen-vl-max-2025-01-25\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2.0-pro": {
"id": "doubao-seed-2.0-pro",
"name": "Doubao Seed 2.0 Pro",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/doubao-seed-2.0-pro\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2.0-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-max": {
"id": "qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-24",
"last_updated": "2025-09-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-max\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.5": {
"id": "glm-4.5",
"name": "GLM 4.5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 98304
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/glm-4.5\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-3.5-haiku": {
"id": "claude-3.5-haiku",
"name": "Claude 3.5 Haiku",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-26",
"last_updated": "2025-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/claude-3.5-haiku\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"claude-3.5-haiku\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-1.6-thinking": {
"id": "doubao-seed-1.6-thinking",
"name": "Doubao-Seed 1.6 Thinking",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-15",
"last_updated": "2025-08-15",
"modalities": {
"input": [
"image",
"text",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/doubao-seed-1.6-thinking\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-1.6-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-1.5-vision-pro": {
"id": "doubao-1.5-vision-pro",
"name": "Doubao 1.5 Vision Pro",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/doubao-1.5-vision-pro\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"doubao-1.5-vision-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-1.6-flash": {
"id": "doubao-seed-1.6-flash",
"name": "Doubao-Seed 1.6 Flash",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-15",
"last_updated": "2025-08-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/doubao-seed-1.6-flash\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-1.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M1": {
"id": "MiniMax-M1",
"name": "MiniMax M1",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 80000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/MiniMax-M1\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-30b-a3b-instruct-2507": {
"id": "qwen3-30b-a3b-instruct-2507",
"name": "Qwen3 30b A3b Instruct 2507",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-04",
"last_updated": "2026-02-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-30b-a3b-instruct-2507\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-30b-a3b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-30b-a3b": {
"id": "qwen3-30b-a3b",
"name": "Qwen3 30B A3B",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 40000,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-30b-a3b\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-30b-a3b-thinking": {
"id": "qwen3-vl-30b-a3b-thinking",
"name": "Qwen3-Vl 30b A3b Thinking",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-09",
"last_updated": "2026-02-09",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-vl-30b-a3b-thinking\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-30b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2": {
"id": "kimi-k2",
"name": "Kimi K2",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/kimi-k2\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-3.7-sonnet": {
"id": "claude-3.7-sonnet",
"name": "Claude 3.7 Sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/claude-3.7-sonnet\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"claude-3.7-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b": {
"id": "qwen3-235b-a22b",
"name": "Qwen 3 235B A22B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-235b-a22b\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-instruct-2507": {
"id": "qwen3-235b-a22b-instruct-2507",
"name": "Qwen3 235b A22B Instruct 2507",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-12",
"last_updated": "2025-08-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 64000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-235b-a22b-instruct-2507\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3-0324": {
"id": "deepseek-v3-0324",
"name": "DeepSeek-V3-0324",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/deepseek-v3-0324\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3-0324\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-1.5-thinking-pro": {
"id": "doubao-1.5-thinking-pro",
"name": "Doubao 1.5 Thinking Pro",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/doubao-1.5-thinking-pro\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"doubao-1.5-thinking-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.0-flash-preview": {
"id": "gemini-3.0-flash-preview",
"name": "Gemini 3.0 Flash Preview",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-18",
"last_updated": "2025-12-18",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/gemini-3.0-flash-preview\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.0-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-4.5-sonnet": {
"id": "claude-4.5-sonnet",
"name": "Claude 4.5 Sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/claude-4.5-sonnet\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"claude-4.5-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "gpt-oss-120b",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-06",
"last_updated": "2025-08-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/gpt-oss-120b\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-max-2025-01-25": {
"id": "qwen-max-2025-01-25",
"name": "Qwen2.5-Max-2025-01-25",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen-max-2025-01-25\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen-max-2025-01-25\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-480b-a35b-instruct": {
"id": "qwen3-coder-480b-a35b-instruct",
"name": "Qwen3 Coder 480B A35B Instruct",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-14",
"last_updated": "2025-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/qwen3-coder-480b-a35b-instruct\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-480b-a35b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/gemini-2.5-pro\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2.0-lite": {
"id": "doubao-seed-2.0-lite",
"name": "Doubao Seed 2.0 Lite",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/doubao-seed-2.0-lite\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2.0-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-4.5-haiku": {
"id": "claude-4.5-haiku",
"name": "Claude 4.5 Haiku",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-10-16",
"last_updated": "2025-10-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/claude-4.5-haiku\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"claude-4.5-haiku\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 64000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/gemini-2.5-flash\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-4.0-sonnet": {
"id": "claude-4.0-sonnet",
"name": "Claude 4.0 Sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/claude-4.0-sonnet\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"claude-4.0-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3.1": {
"id": "deepseek-v3.1",
"name": "DeepSeek-V3.1",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-19",
"last_updated": "2025-08-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/deepseek-v3.1\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun-ai/gelab-zero-4b-preview": {
"id": "stepfun-ai/gelab-zero-4b-preview",
"name": "Stepfun-Ai/Gelab Zero 4b Preview",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/stepfun-ai/gelab-zero-4b-preview\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"stepfun-ai/gelab-zero-4b-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meituan/longcat-flash-lite": {
"id": "meituan/longcat-flash-lite",
"name": "Meituan/Longcat-Flash-Lite",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-06",
"last_updated": "2026-02-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 320000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/meituan/longcat-flash-lite\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"meituan/longcat-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meituan/longcat-flash-chat": {
"id": "meituan/longcat-flash-chat",
"name": "Meituan/Longcat-Flash-Chat",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-11-05",
"last_updated": "2025-11-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/meituan/longcat-flash-chat\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"meituan/longcat-flash-chat\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun/step-3.5-flash": {
"id": "stepfun/step-3.5-flash",
"name": "Stepfun/Step-3.5 Flash",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-02",
"last_updated": "2026-02-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 64000,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/stepfun/step-3.5-flash\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"stepfun/step-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2-flash": {
"id": "xiaomi/mimo-v2-flash",
"name": "Xiaomi/Mimo-V2-Flash",
"description": "MiMo flash model for fast multimodal assistance and agent workflows",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12-01",
"release_date": "2025-12-16",
"last_updated": "2026-02-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/xiaomi/mimo-v2-flash\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.1": {
"id": "minimax/minimax-m2.1",
"name": "Minimax/Minimax-M2.1",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 204800,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/minimax/minimax-m2.1\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2": {
"id": "minimax/minimax-m2",
"name": "Minimax/Minimax-M2",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-10-28",
"last_updated": "2025-10-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/minimax/minimax-m2\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5": {
"id": "minimax/minimax-m2.5",
"name": "Minimax/Minimax-M2.5",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 204800,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/minimax/minimax-m2.5\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5-highspeed": {
"id": "minimax/minimax-m2.5-highspeed",
"name": "Minimax/Minimax-M2.5 Highspeed",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 204800,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/minimax/minimax-m2.5-highspeed\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-math-v2": {
"id": "deepseek/deepseek-math-v2",
"name": "Deepseek/Deepseek-Math-V2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-04",
"last_updated": "2025-12-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 160000,
"output": 160000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/deepseek/deepseek-math-v2\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-math-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.2-exp-thinking": {
"id": "deepseek/deepseek-v3.2-exp-thinking",
"name": "DeepSeek/DeepSeek-V3.2-Exp-Thinking",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/deepseek/deepseek-v3.2-exp-thinking\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.2-exp-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.1-terminus-thinking": {
"id": "deepseek/deepseek-v3.1-terminus-thinking",
"name": "DeepSeek/DeepSeek-V3.1-Terminus-Thinking",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-22",
"last_updated": "2025-09-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/deepseek/deepseek-v3.1-terminus-thinking\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.1-terminus-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.2-exp": {
"id": "deepseek/deepseek-v3.2-exp",
"name": "DeepSeek/DeepSeek-V3.2-Exp",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/deepseek/deepseek-v3.2-exp\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.2-exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.1-terminus": {
"id": "deepseek/deepseek-v3.1-terminus",
"name": "DeepSeek/DeepSeek-V3.1-Terminus",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-22",
"last_updated": "2025-09-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/deepseek/deepseek-v3.1-terminus\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.1-terminus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.2-251201": {
"id": "deepseek/deepseek-v3.2-251201",
"name": "Deepseek/DeepSeek-V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/deepseek/deepseek-v3.2-251201\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.2-251201\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-code-fast-1": {
"id": "x-ai/grok-code-fast-1",
"name": "x-AI/Grok-Code-Fast 1",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-02",
"last_updated": "2025-09-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 10000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/x-ai/grok-code-fast-1\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-code-fast-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4-fast-reasoning": {
"id": "x-ai/grok-4-fast-reasoning",
"name": "X-Ai/Grok-4-Fast-Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-18",
"last_updated": "2025-12-18",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/x-ai/grok-4-fast-reasoning\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.1-fast-non-reasoning": {
"id": "x-ai/grok-4.1-fast-non-reasoning",
"name": "X-Ai/Grok 4.1 Fast Non Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-19",
"last_updated": "2025-12-19",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/x-ai/grok-4.1-fast-non-reasoning\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.1-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.1-fast-reasoning": {
"id": "x-ai/grok-4.1-fast-reasoning",
"name": "X-Ai/Grok 4.1 Fast Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-19",
"last_updated": "2025-12-19",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 20000000,
"output": 2000000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/x-ai/grok-4.1-fast-reasoning\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.1-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4-fast": {
"id": "x-ai/grok-4-fast",
"name": "x-AI/Grok-4-Fast",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-20",
"last_updated": "2025-09-20",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/x-ai/grok-4-fast\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4-fast-non-reasoning": {
"id": "x-ai/grok-4-fast-non-reasoning",
"name": "X-Ai/Grok-4-Fast-Non-Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-18",
"last_updated": "2025-12-18",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/x-ai/grok-4-fast-non-reasoning\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.1-fast": {
"id": "x-ai/grok-4.1-fast",
"name": "x-AI/Grok-4.1-Fast",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-11-20",
"last_updated": "2025-11-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/x-ai/grok-4.1-fast\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.1-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2": {
"id": "openai/gpt-5.2",
"name": "OpenAI/GPT-5.2",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/openai/gpt-5.2\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5": {
"id": "openai/gpt-5",
"name": "OpenAI/GPT-5",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-19",
"last_updated": "2025-09-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/openai/gpt-5\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-0905": {
"id": "moonshotai/kimi-k2-0905",
"name": "Kimi K2 0905",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-08",
"last_updated": "2025-09-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 100000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/moonshotai/kimi-k2-0905\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-0905\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-thinking": {
"id": "moonshotai/kimi-k2-thinking",
"name": "Kimi K2 Thinking",
"description": "Kimi reasoning model for long-horizon research, planning, and tool use",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-11-07",
"last_updated": "2025-11-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 100000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/moonshotai/kimi-k2-thinking\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.5": {
"id": "moonshotai/kimi-k2.5",
"name": "Moonshotai/Kimi-K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-01-28",
"last_updated": "2026-01-28",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/moonshotai/kimi-k2.5\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7": {
"id": "z-ai/glm-4.7",
"name": "Z-Ai/GLM 4.7",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 200000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/z-ai/glm-4.7\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.6": {
"id": "z-ai/glm-4.6",
"name": "Z-AI/GLM 4.6",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-10-11",
"last_updated": "2025-10-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 200000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/z-ai/glm-4.6\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/autoglm-phone-9b": {
"id": "z-ai/autoglm-phone-9b",
"name": "Z-Ai/Autoglm Phone 9b",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 12800,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/z-ai/autoglm-phone-9b\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/autoglm-phone-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5": {
"id": "z-ai/glm-5",
"name": "Z-Ai/GLM 5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qiniu-ai/z-ai/glm-5\", apiKey: processEnvironment[\"QINIU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.qnaigc.com/v1\")!,\n apiKey: processEnvironment[\"QINIU_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"ambient": {
"id": "ambient",
"name": "Ambient",
"baseURL": "https://api.ambient.xyz/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"AMBIENT_API_KEY"
],
"doc": "https://ambient.xyz",
"modelCount": 10,
"models": {
"ambient/large": {
"id": "ambient/large",
"name": "Ambient Large",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 202752
},
"cost": {
"input": 0.6,
"output": 2,
"cache_read": 0.15,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ambient/ambient/large\", apiKey: processEnvironment[\"AMBIENT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.ambient.xyz/v1\")!,\n apiKey: processEnvironment[\"AMBIENT_API_KEY\"]\n)\nlet session = provider.model(\"ambient/large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun/step-3.7-flash": {
"id": "stepfun/step-3.7-flash",
"name": "Step 3.7 Flash",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.19,
"output": 1.14,
"cache_read": 0.03,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ambient/stepfun/step-3.7-flash\", apiKey: processEnvironment[\"AMBIENT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.ambient.xyz/v1\")!,\n apiKey: processEnvironment[\"AMBIENT_API_KEY\"]\n)\nlet session = provider.model(\"stepfun/step-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5": {
"id": "xiaomi/mimo-v2.5",
"name": "MiMo-V2.5",
"description": "MiMo omni model for text, image, video, audio, and agents",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.4,
"output": 2,
"cache_read": 0.08,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ambient/xiaomi/mimo-v2.5\", apiKey: processEnvironment[\"AMBIENT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.ambient.xyz/v1\")!,\n apiKey: processEnvironment[\"AMBIENT_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2-FP8": {
"id": "zai-org/GLM-5.2-FP8",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 202752
},
"cost": {
"input": 1.2,
"output": 4.2,
"cache_read": 0.26,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ambient/zai-org/GLM-5.2-FP8\", apiKey: processEnvironment[\"AMBIENT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.ambient.xyz/v1\")!,\n apiKey: processEnvironment[\"AMBIENT_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.1-FP8": {
"id": "zai-org/GLM-5.1-FP8",
"name": "GLM 5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ambient/zai-org/GLM-5.1-FP8\", apiKey: processEnvironment[\"AMBIENT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.ambient.xyz/v1\")!,\n apiKey: processEnvironment[\"AMBIENT_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.1-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-0731": {
"id": "deepseek/deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.08,
"output": 0.18,
"cache_read": 0.016,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ambient/deepseek/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"AMBIENT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.ambient.xyz/v1\")!,\n apiKey: processEnvironment[\"AMBIENT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash": {
"id": "deepseek/deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.028,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ambient/deepseek/deepseek-v4-flash\", apiKey: processEnvironment[\"AMBIENT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.ambient.xyz/v1\")!,\n apiKey: processEnvironment[\"AMBIENT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.6": {
"id": "moonshotai/kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.2,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ambient/moonshotai/kimi-k2.6\", apiKey: processEnvironment[\"AMBIENT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.ambient.xyz/v1\")!,\n apiKey: processEnvironment[\"AMBIENT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.7-code": {
"id": "moonshotai/kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.69,
"output": 3.49,
"cache_read": 0.14,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ambient/moonshotai/kimi-k2.7-code\", apiKey: processEnvironment[\"AMBIENT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.ambient.xyz/v1\")!,\n apiKey: processEnvironment[\"AMBIENT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.2": {
"id": "z-ai/glm-5.2",
"name": "GLM-5.2",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 202752
},
"cost": {
"input": 0.6,
"output": 2,
"cache_read": 0.15,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ambient/z-ai/glm-5.2\", apiKey: processEnvironment[\"AMBIENT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.ambient.xyz/v1\")!,\n apiKey: processEnvironment[\"AMBIENT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"agentrouter": {
"id": "agentrouter",
"name": "AgentRouter",
"baseURL": "https://agentrouter.org/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"AGENTROUTER_API_KEY"
],
"doc": "https://agentrouter.org/docs/opencode.html",
"modelCount": 3,
"models": {
"gpt-5.6-sol": {
"id": "gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"agentrouter/gpt-5.6-sol\", apiKey: processEnvironment[\"AGENTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://agentrouter.org/v1\")!,\n apiKey: processEnvironment[\"AGENTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5": {
"id": "claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://agentrouter.org/v1"
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"agentrouter/claude-opus-5\", apiKey: processEnvironment[\"AGENTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://agentrouter.org/v1\")!,\n apiKey: processEnvironment[\"AGENTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://agentrouter.org/v1"
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"agentrouter/claude-opus-4-8\", apiKey: processEnvironment[\"AGENTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://agentrouter.org/v1\")!,\n apiKey: processEnvironment[\"AGENTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"xiaomi-token-plan-cn": {
"id": "xiaomi-token-plan-cn",
"name": "Xiaomi Token Plan (China)",
"baseURL": "https://token-plan-cn.xiaomimimo.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"XIAOMI_API_KEY"
],
"doc": "https://platform.xiaomimimo.com/#/docs",
"modelCount": 7,
"models": {
"mimo-v2.5-tts": {
"id": "mimo-v2.5-tts",
"name": "MiMo-V2.5-TTS",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "mimo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-cn/mimo-v2.5-tts\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-cn.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-tts-voiceclone": {
"id": "mimo-v2.5-tts-voiceclone",
"name": "MiMo-V2.5-TTS-VoiceClone",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "mimo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-cn/mimo-v2.5-tts-voiceclone\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-cn.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-tts-voiceclone\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-tts": {
"id": "mimo-v2-tts",
"name": "MiMo-V2-TTS",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "mimo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-cn/mimo-v2-tts\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-cn.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-tts-voicedesign": {
"id": "mimo-v2.5-tts-voicedesign",
"name": "MiMo-V2.5-TTS-VoiceDesign",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "mimo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-cn/mimo-v2.5-tts-voicedesign\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-cn.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-tts-voicedesign\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-pro": {
"id": "mimo-v2-pro",
"name": "MiMo-V2-Pro",
"description": "Earlier MiMo Pro model for multimodal agents, reasoning, and code tasks",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"status": "deprecated",
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-cn/mimo-v2-pro\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-cn.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5": {
"id": "mimo-v2.5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-cn/mimo-v2.5\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-cn.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-pro": {
"id": "mimo-v2.5-pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-cn/mimo-v2.5-pro\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-cn.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"nano-gpt": {
"id": "nano-gpt",
"name": "NanoGPT",
"baseURL": "https://nano-gpt.com/api/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"NANO_GPT_API_KEY"
],
"doc": "https://docs.nano-gpt.com",
"modelCount": 594,
"models": {
"glm-4.1v-thinking-flashx": {
"id": "glm-4.1v-thinking-flashx",
"name": "GLM 4.1V Thinking FlashX",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "glm",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-09",
"last_updated": "2025-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 64000,
"input": 64000,
"output": 8192
},
"cost": {
"input": 0.3,
"output": 0.3,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/glm-4.1v-thinking-flashx\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.1v-thinking-flashx\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-thinking:8192": {
"id": "claude-opus-4-thinking:8192",
"name": "Claude 4 Opus Thinking (8K)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-thinking:8192\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-thinking:8192\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-max": {
"id": "qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.5,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.7-max\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"longcat-2.0": {
"id": "longcat-2.0",
"name": "LongCat 2.0",
"description": "Meituan LongCat-2.0, a reasoning model with tool calling and a 1M-token context window",
"family": "longcat",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048756,
"input": 1048756,
"output": 262144
},
"cost": {
"input": 0.75,
"output": 3,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/longcat-2.0\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"longcat-2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it-garnet": {
"id": "gemma-4-31b-it-garnet",
"name": "Garnet",
"description": "Garnet is a multimodal Gemma 4 31B creative finetune for expressive dialogue, long-form storytelling, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.45,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-31b-it-garnet\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it-garnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"exa-answer": {
"id": "exa-answer",
"name": "Exa (Answer)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-23",
"last_updated": "2025-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 4096,
"input": 4096,
"output": 4096
},
"cost": {
"input": 2.5,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/exa-answer\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"exa-answer\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.0-pro-exp-02-05": {
"id": "gemini-2.0-pro-exp-02-05",
"name": "Gemini 2.0 Pro 0205",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-02-05",
"last_updated": "2025-02-05",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2097152,
"input": 2097152,
"output": 8192
},
"cost": {
"input": 1.989,
"output": 7.956,
"cache_read": 0.49725
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.0-pro-exp-02-05\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.0-pro-exp-02-05\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Meta-Llama-3-1-8B-Instruct-FP8": {
"id": "Meta-Llama-3-1-8B-Instruct-FP8",
"name": "Llama 3.1 8B (decentralized)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 0.02,
"output": 0.03,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Meta-Llama-3-1-8B-Instruct-FP8\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Meta-Llama-3-1-8B-Instruct-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2": {
"id": "MiniMax-M2",
"name": "MiniMax M2",
"description": "Efficient open MiniMax model built for coding agents and tool-heavy workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-10-27",
"last_updated": "2025-10-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 131072
},
"cost": {
"input": 0.17,
"output": 1.53,
"cache_read": 0.085
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/MiniMax-M2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ernie-5.0-thinking-preview": {
"id": "ernie-5.0-thinking-preview",
"name": "Ernie 5.0 Thinking Preview",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "ernie",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"release_date": "2025-11-18",
"last_updated": "2025-11-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 1,
"output": 3.5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/ernie-5.0-thinking-preview\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"ernie-5.0-thinking-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mercury-coder-small": {
"id": "mercury-coder-small",
"name": "Mercury Coder Small",
"description": "Model by Inception AI. A diffusion large language model that runs incredibly quickly (500+ tokens/second) while matching Claude 3.5 Haiku and GPT-4o-mini. 1st in speed on Copilot arena, and matching 2nd in quality.",
"family": "mercury",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 0.25,
"output": 1,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mercury-coder-small\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mercury-coder-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-it-luminous": {
"id": "gemma-4-26b-a4b-it-luminous",
"name": "Luminous Mirror",
"description": "Luminous Mirror is a Gemma 4 26B A4B multimodal mixture-of-experts fine-tune for creative writing, expressive dialogue, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-26b-a4b-it-luminous\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it-luminous\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-thinking:32768": {
"id": "claude-sonnet-4-thinking:32768",
"name": "Claude 4 Sonnet Thinking (32K)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-sonnet-4-thinking:32768\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-thinking:32768\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-it-shadowsiren": {
"id": "gemma-4-26b-a4b-it-shadowsiren",
"name": "Shadow Siren",
"description": "Shadow Siren is a Gemma 4 26B A4B multimodal mixture-of-experts fine-tune for creative writing, expressive dialogue, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-26b-a4b-it-shadowsiren\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it-shadowsiren\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"auto-model-premium": {
"id": "auto-model-premium",
"name": "Auto model (Premium)",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"family": "auto",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2025-04-16",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 1000000
},
"cost": {
"input": 9.996,
"output": 19.992,
"cache_read": 4.998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/auto-model-premium\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"auto-model-premium\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-code-latest": {
"id": "mistral-code-latest",
"name": "Mistral Code Latest",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "mistral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 0.9,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistral-code-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistral-code-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-122b-a10b:thinking": {
"id": "qwen3.5-122b-a10b:thinking",
"name": "Qwen3.5 122B A10B Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 81920
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 0.437,
"output": 3.496,
"cache_read": 0.103788
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-122b-a10b:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-122b-a10b:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-1-6-250615": {
"id": "doubao-seed-1-6-250615",
"name": "Doubao Seed 1.6",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-06-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 16384
},
"cost": {
"input": 0.204,
"output": 0.51,
"cache_read": 0.102
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/doubao-seed-1-6-250615\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-1-6-250615\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Gemma-4-26B-A4B-MeroMero": {
"id": "Gemma-4-26B-A4B-MeroMero",
"name": "Gemma 4 26B A4B MeroMero",
"description": "Gemma 4 26B A4B MeroMero is an NVFP4 multimodal mixture-of-experts fine-tune for emotive dialogue, relationship scenes, creative writing, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Gemma-4-26B-A4B-MeroMero\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Gemma-4-26B-A4B-MeroMero\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-max": {
"id": "qwen-max",
"name": "Qwen 2.5 Max",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-04-03",
"last_updated": "2025-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"input": 32000,
"output": 8192
},
"cost": {
"input": 1.5997,
"output": 6.392,
"cache_read": 0.79985
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen-max\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claw-low": {
"id": "claw-low",
"name": "Claw Low",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-11",
"last_updated": "2026-05-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claw-low\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claw-low\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-0-mini-260215": {
"id": "doubao-seed-2-0-mini-260215",
"name": "Doubao Seed 2.0 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 32000
},
"cost": {
"input": 0.0493,
"output": 0.4845,
"cache_read": 0.02465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/doubao-seed-2-0-mini-260215\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-mini-260215\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-27b": {
"id": "qwen3.5-27b",
"name": "Qwen3.5 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 260096,
"input": 260096,
"output": 65536
},
"cost": {
"input": 0.27,
"output": 2.16,
"cache_read": 0.135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-27b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1-thinking:1024": {
"id": "claude-opus-4-1-thinking:1024",
"name": "Claude 4.1 Opus Thinking (1K)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-1-thinking:1024\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-1-thinking:1024\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it-gemsicle": {
"id": "gemma-4-31b-it-gemsicle",
"name": "Gemsicle",
"description": "Gemsicle is a multimodal Gemma 4 31B creative finetune for expressive dialogue, long-form storytelling, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.45,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-31b-it-gemsicle\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it-gemsicle\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-27b": {
"id": "qwen3.8-27b",
"name": "Qwen3.8 27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.7,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.8-27b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-35b-a3b": {
"id": "qwen3.5-35b-a3b",
"name": "Qwen3.5 35B A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 260096,
"input": 260096,
"output": 65536
},
"cost": {
"input": 0.225,
"output": 1.8,
"cache_read": 0.1125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-35b-a3b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5-20251001-thinking": {
"id": "claude-haiku-4-5-20251001-thinking",
"name": "Claude Haiku 4.5 Thinking",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-haiku-4-5-20251001-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5-20251001-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-preview-09-2025-thinking": {
"id": "gemini-2.5-flash-preview-09-2025-thinking",
"name": "Gemini 2.5 Flash Preview (09/2025) – Thinking",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text",
"image",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash-preview-09-2025-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-preview-09-2025-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-it-opusdistill": {
"id": "gemma-4-26b-a4b-it-opusdistill",
"name": "Opus Distill",
"description": "Opus Distill is a Gemma 4 26B A4B multimodal mixture-of-experts fine-tune for creative writing, expressive dialogue, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-26b-a4b-it-opusdistill\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it-opusdistill\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4-plus-0111": {
"id": "glm-4-plus-0111",
"name": "GLM 4 Plus 0111",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-02-19",
"last_updated": "2025-02-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 4096
},
"cost": {
"input": 9.996,
"output": 9.996,
"cache_read": 4.998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/glm-4-plus-0111\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-4-plus-0111\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.0-pro-reasoner": {
"id": "gemini-2.0-pro-reasoner",
"name": "Gemini 2.0 Pro Reasoner",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-02-05",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 1.292,
"output": 4.998,
"cache_read": 0.323
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.0-pro-reasoner\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.0-pro-reasoner\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3.5-27B-Queen-Derestricted": {
"id": "Qwen3.5-27B-Queen-Derestricted",
"name": "Qwen3.5 27B Queen Derestricted",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-04-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.306,
"output": 0.306,
"cache_read": 0.153
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Qwen3.5-27B-Queen-Derestricted\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3.5-27B-Queen-Derestricted\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-3.6-plus": {
"id": "qwen-3.6-plus",
"name": "Qwen 3.6 Plus",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "qwen3.6",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991808,
"input": 991808,
"output": 65536
},
"cost": {
"input": 0.325,
"output": 1.95,
"cache_read": 0.0325,
"cache_write": 0.40625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen-3.6-plus\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen-3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Gemma-4-31B-Cognitive-Unshackled": {
"id": "Gemma-4-31B-Cognitive-Unshackled",
"name": "Gemma 4 31B Cognitive Unshackled",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-05-01",
"last_updated": "2026-05-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.306,
"output": 0.306,
"cache_read": 0.153
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Gemma-4-31B-Cognitive-Unshackled\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Gemma-4-31B-Cognitive-Unshackled\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro-preview-06-05": {
"id": "gemini-2.5-pro-preview-06-05",
"name": "Gemini 2.5 Pro Preview 0605",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"release_date": "2025-06-05",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-pro-preview-06-05\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro-preview-06-05\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-thinking:8192": {
"id": "claude-sonnet-4-thinking:8192",
"name": "Claude 4 Sonnet Thinking (8K)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-sonnet-4-thinking:8192\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-thinking:8192\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-turbo": {
"id": "qwen-turbo",
"name": "Qwen Turbo",
"description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-11-01",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 8192
},
"cost": {
"input": 0.04998,
"output": 0.2006,
"cache_read": 0.02499
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen-turbo\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-flash:thinking": {
"id": "qwen3.7-flash:thinking",
"name": "Qwen3.7 Flash Thinking",
"description": "Lightweight multimodal Qwen model for high-throughput text, image, and video tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 983616,
"input": 983616,
"output": 65536
},
"cost": {
"input": 0.03,
"output": 0.13,
"cache_read": 0.006,
"cache_write": 0.038
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.7-flash:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-flash:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-omni-plus": {
"id": "qwen3.5-omni-plus",
"name": "Qwen3.5 Omni Plus",
"description": "Omni-modal model for text, vision, audio, and multimodal agent tasks",
"family": "qwen3.5",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-03-30",
"last_updated": "2026-03-30",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 983616,
"input": 983616,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-omni-plus\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-omni-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite": {
"id": "gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash Lite",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"asi1-mini": {
"id": "asi1-mini",
"name": "ASI1 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-03-25",
"last_updated": "2025-03-25",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 1,
"output": 1,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/asi1-mini\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"asi1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro-preview-03-25": {
"id": "gemini-2.5-pro-preview-03-25",
"name": "Gemini 2.5 Pro Preview 0325",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"release_date": "2025-03-25",
"last_updated": "2025-03-25",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-pro-preview-03-25\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro-preview-03-25\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5-20250929-thinking": {
"id": "claude-sonnet-4-5-20250929-thinking",
"name": "Claude Sonnet 4.5 Thinking",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-sonnet-4-5-20250929-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5-20250929-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4-air-0111": {
"id": "glm-4-air-0111",
"name": "GLM 4 Air 0111",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-01-11",
"last_updated": "2025-01-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 4096
},
"cost": {
"input": 0.1394,
"output": 0.1394,
"cache_read": 0.0697
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/glm-4-air-0111\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-4-air-0111\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1-20250805": {
"id": "claude-opus-4-1-20250805",
"name": "Claude 4.1 Opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-1-20250805\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-1-20250805\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen25-vl-72b-instruct": {
"id": "qwen25-vl-72b-instruct",
"name": "Qwen25 VL 72b",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-05-10",
"last_updated": "2025-05-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"input": 32000,
"output": 115200
},
"cost": {
"input": 0.69989,
"output": 0.69989,
"cache_read": 0.349945
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen25-vl-72b-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen25-vl-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-flash:thinking": {
"id": "qwen3.5-flash:thinking",
"name": "Qwen3.5 Flash Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 81920
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991808,
"input": 991808,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-flash:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-flash:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"command-a-reasoning-08-2025": {
"id": "command-a-reasoning-08-2025",
"name": "Cohere Command A (08/2025)",
"description": "Cohere reasoning model for multilingual enterprise agents, tools, and complex workflows",
"family": "command-a",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-06-01",
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 8192
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/command-a-reasoning-08-2025\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"command-a-reasoning-08-2025\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"phi-4-multimodal-instruct": {
"id": "phi-4-multimodal-instruct",
"name": "Phi 4 Multimodal",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "phi",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-26",
"last_updated": "2025-07-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 0.07,
"output": 0.11,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/phi-4-multimodal-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"phi-4-multimodal-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-thinking:32768": {
"id": "claude-opus-4-thinking:32768",
"name": "Claude 4 Opus Thinking (32K)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-thinking:32768\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-thinking:32768\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-flash": {
"id": "qwen3.5-flash",
"name": "Qwen3.5 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991808,
"input": 991808,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-code-agent-latest": {
"id": "mistral-code-agent-latest",
"name": "Mistral Code Agent Latest",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "mistral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistral-code-agent-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistral-code-agent-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3.5-27B-BlueStar-v3-Derestricted": {
"id": "Qwen3.5-27B-BlueStar-v3-Derestricted",
"name": "Qwen3.5 27B BlueStar v3 Derestricted",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-04-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.306,
"output": 0.306,
"cache_read": 0.153
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Qwen3.5-27B-BlueStar-v3-Derestricted\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3.5-27B-BlueStar-v3-Derestricted\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-reasoner": {
"id": "deepseek-reasoner",
"name": "DeepSeek Reasoner",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-12-01",
"last_updated": "2026-02-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 64000,
"input": 64000,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 1.7,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-reasoner\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-reasoner\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-e4b-it": {
"id": "gemma-4-e4b-it",
"name": "Gemma 4 E4B Instruct",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 16384
},
"cost": {
"input": 0.04,
"output": 0.2,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-e4b-it\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-e4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"GLM-4.6-Derestricted-v5": {
"id": "GLM-4.6-Derestricted-v5",
"name": "GLM 4.6 Derestricted v5",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 8192
},
"cost": {
"input": 0.4,
"output": 1.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/GLM-4.6-Derestricted-v5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"GLM-4.6-Derestricted-v5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-1-6-flash-250615": {
"id": "doubao-seed-1-6-flash-250615",
"name": "Doubao Seed 1.6 Flash",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-06-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 16384
},
"cost": {
"input": 0.0374,
"output": 0.374,
"cache_read": 0.0187
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/doubao-seed-1-6-flash-250615\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-1-6-flash-250615\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite-preview-06-17": {
"id": "gemini-2.5-flash-lite-preview-06-17",
"name": "Gemini 2.5 Flash Lite Preview",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash-lite-preview-06-17\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite-preview-06-17\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-thinking": {
"id": "claude-opus-4-thinking",
"name": "Claude 4 Opus Thinking",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"brave-research": {
"id": "brave-research",
"name": "Brave (Research)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-02-10",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"input": 16384,
"output": 16384
},
"cost": {
"input": 5,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/brave-research\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"brave-research\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-1.5-pro-256k": {
"id": "doubao-1.5-pro-256k",
"name": "Doubao 1.5 Pro 256k",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-03-12",
"last_updated": "2025-03-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 16384
},
"cost": {
"input": 0.799,
"output": 1.445,
"cache_read": 0.3995
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/doubao-1.5-pro-256k\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"doubao-1.5-pro-256k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-30b-a3b-instruct": {
"id": "qwen3-coder-30b-a3b-instruct",
"name": "Qwen3 Coder 30B A3B Instruct",
"description": "Smaller Qwen coder for efficient local agents and repo-level fixes",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-z1-airx": {
"id": "glm-z1-airx",
"name": "GLM Z1 AirX",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2025-04-15",
"last_updated": "2025-04-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"input": 32000,
"output": 16384
},
"cost": {
"input": 0.7,
"output": 0.7,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/glm-z1-airx\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-z1-airx\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ernie-5.1:thinking": {
"id": "ernie-5.1:thinking",
"name": "ERNIE 5.1 Thinking",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "ernie",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"release_date": "2026-05-10",
"last_updated": "2026-05-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 119000,
"input": 119000,
"output": 64000
},
"cost": {
"input": 0.75,
"output": 3,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/ernie-5.1:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"ernie-5.1:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max:thinking": {
"id": "qwen3.8-max:thinking",
"name": "Qwen3.8 Max Thinking",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"input": 991000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.8-max:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-max:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"universal-summarizer": {
"id": "universal-summarizer",
"name": "Universal Summarizer",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-23",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 30,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/universal-summarizer\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"universal-summarizer\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"venice-uncensored": {
"id": "venice-uncensored",
"name": "Venice Uncensored",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "venice",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-10-01",
"last_updated": "2025-02-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 8192
},
"cost": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/venice-uncensored\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"venice-uncensored\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-preview-09-2025": {
"id": "gemini-2.5-flash-preview-09-2025",
"name": "Gemini 2.5 Flash Preview (09/2025)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text",
"image",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash-preview-09-2025\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-preview-09-2025\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-20250514": {
"id": "claude-opus-4-20250514",
"name": "Claude 4 Opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-20250514\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-20250514\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-1.5-pro-32k": {
"id": "doubao-1.5-pro-32k",
"name": "Doubao 1.5 Pro 32k",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-11-20",
"last_updated": "2025-01-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"input": 32000,
"output": 8192
},
"cost": {
"input": 0.1343,
"output": 0.3349,
"cache_read": 0.06715
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/doubao-1.5-pro-32k\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"doubao-1.5-pro-32k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-it-moonlight": {
"id": "gemma-4-26b-a4b-it-moonlight",
"name": "Moonlight Dusk",
"description": "Moonlight Dusk is a Gemma 4 26B A4B multimodal mixture-of-experts fine-tune for creative writing, expressive dialogue, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-26b-a4b-it-moonlight\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it-moonlight\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-chat-cheaper": {
"id": "deepseek-chat-cheaper",
"name": "DeepSeek V3/Chat Cheaper",
"description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
"family": "deepseek",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2025-04-15",
"last_updated": "2025-04-15",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 8192
},
"cost": {
"input": 0.1,
"output": 0.425,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-chat-cheaper\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-chat-cheaper\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-it-darksoul": {
"id": "gemma-4-26b-a4b-it-darksoul",
"name": "Dark Soul",
"description": "Dark Soul is a Gemma 4 26B A4B multimodal mixture-of-experts fine-tune for creative writing, expressive dialogue, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-26b-a4b-it-darksoul\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it-darksoul\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite-preview-09-2025": {
"id": "gemini-2.5-flash-lite-preview-09-2025",
"name": "Gemini 2.5 Flash Lite Preview (09/2025)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text",
"image",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash-lite-preview-09-2025\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite-preview-09-2025\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Gemma-4-31B-Queen": {
"id": "Gemma-4-31B-Queen",
"name": "Gemma 4 31B Queen",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-05-01",
"last_updated": "2026-05-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.306,
"output": 0.306,
"cache_read": 0.153
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Gemma-4-31B-Queen\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Gemma-4-31B-Queen\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1-sambanova": {
"id": "deepseek-r1-sambanova",
"name": "DeepSeek R1 Fast",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-02-20",
"last_updated": "2025-02-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 4096
},
"cost": {
"input": 4.998,
"output": 6.987,
"cache_read": 2.499
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-r1-sambanova\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1-sambanova\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1": {
"id": "deepseek-r1",
"name": "DeepSeek R1",
"description": "Classic open reasoning model for transparent math, coding, and deliberate problem solving",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-01-20",
"last_updated": "2025-05-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 8192
},
"cost": {
"input": 0.4,
"output": 1.7,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-r1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity-academic-researcher": {
"id": "perplexity-academic-researcher",
"name": "Perplexity Academic Researcher",
"description": "Sonar Reasoning Pro with Perplexity's academic search mode. Prioritizes scholarly and peer-reviewed sources from academic repositories and returns cited research synthesis.",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"release_date": "2026-07-10",
"last_updated": "2026-07-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 115200
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/perplexity-academic-researcher\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"perplexity-academic-researcher\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-thinking:64000": {
"id": "claude-sonnet-4-thinking:64000",
"name": "Claude 4 Sonnet Thinking (64K)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-sonnet-4-thinking:64000\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-thinking:64000\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-max:thinking": {
"id": "qwen3.7-max:thinking",
"name": "Qwen3.7 Max Thinking",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.5,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.7-max:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-max:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-uncensored": {
"id": "gemma-4-26b-a4b-uncensored",
"name": "Gemma 4 26B A4B Uncensored",
"description": "Gemma 4 26B A4B Uncensored is an FP8 open-weight multimodal mixture-of-experts model LoRA-tuned for fewer refusals across chat, coding, tool use, and long-context work.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-26b-a4b-uncensored\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-uncensored\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5-20251101:thinking": {
"id": "claude-opus-4-5-20251101:thinking",
"name": "Claude 4.5 Opus Thinking",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-01",
"last_updated": "2025-11-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-5-20251101:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-5-20251101:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-27b:thinking": {
"id": "qwen3.5-27b:thinking",
"name": "Qwen3.5 27B Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 81920
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 260096,
"input": 260096,
"output": 65536
},
"cost": {
"input": 0.27,
"output": 2.16,
"cache_read": 0.135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-27b:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-27b:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-0.8b": {
"id": "qwen3.5-0.8b",
"name": "Qwen3.5 0.8B",
"description": "Qwen3.5 0.8B is a lightweight open-weight multimodal model from Alibaba for fast reasoning, visual understanding, tool use, and JSON output.",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-16",
"last_updated": "2026-08-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.06,
"output": 0.12,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-0.8b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-0.8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-0-code-preview-260215": {
"id": "doubao-seed-2-0-code-preview-260215",
"name": "Doubao Seed 2.0 Code Preview",
"description": "ByteDance Seed coding model for multimodal software engineering and long-running agents",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 128000
},
"cost": {
"input": 0.782,
"output": 3.893,
"cache_read": 0.391
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/doubao-seed-2-0-code-preview-260215\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-code-preview-260215\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ernie-5.1": {
"id": "ernie-5.1",
"name": "ERNIE 5.1",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "ernie",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-05-10",
"last_updated": "2026-05-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 119000,
"input": 119000,
"output": 64000
},
"cost": {
"input": 0.75,
"output": 3,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/ernie-5.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"ernie-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-it-chimerax": {
"id": "gemma-4-26b-a4b-it-chimerax",
"name": "Chimera X",
"description": "Chimera X is a Gemma 4 26B A4B multimodal mixture-of-experts fine-tune for creative writing, expressive dialogue, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-26b-a4b-it-chimerax\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it-chimerax\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5-20250929": {
"id": "claude-sonnet-4-5-20250929",
"name": "Claude Sonnet 4.5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-sonnet-4-5-20250929\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5-20250929\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-flash": {
"id": "qwen3.7-flash",
"name": "Qwen3.7 Flash",
"description": "Lightweight multimodal Qwen model for high-throughput text, image, and video tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991808,
"input": 991808,
"output": 65536
},
"cost": {
"input": 0.03,
"output": 0.13,
"cache_read": 0.006,
"cache_write": 0.038
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.7-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sarvam-105b": {
"id": "sarvam-105b",
"name": "Sarvam 105B",
"description": "Flagship Indian-language reasoning model for enterprise multilingual applications",
"family": "sarvam",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 4096
},
"cost": {
"input": 0.054,
"output": 0.2124,
"cache_read": 0.0336
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/sarvam-105b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"sarvam-105b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-chat": {
"id": "deepseek-chat",
"name": "DeepSeek V3/Deepseek Chat",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-12-01",
"last_updated": "2026-02-28",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 8192
},
"cost": {
"input": 0.1,
"output": 0.425,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-chat\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-chat\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"holo3-35b-a3b": {
"id": "holo3-35b-a3b",
"name": "Holo3-35B-A3B",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"input": 65536,
"output": 8192
},
"cost": {
"input": 0.25,
"output": 1.8,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/holo3-35b-a3b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"holo3-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"hermes-high": {
"id": "hermes-high",
"name": "Hermes High",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "hermes",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-11",
"last_updated": "2026-05-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/hermes-high\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"hermes-high\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claw-high": {
"id": "claw-high",
"name": "Claw High",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-11",
"last_updated": "2026-05-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claw-high\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claw-high\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"holo3-35b-a3b:thinking": {
"id": "holo3-35b-a3b:thinking",
"name": "Holo3-35B-A3B Thinking",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"input": 65536,
"output": 8192
},
"cost": {
"input": 0.25,
"output": 1.8,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/holo3-35b-a3b:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"holo3-35b-a3b:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-omni-flash": {
"id": "qwen3.5-omni-flash",
"name": "Qwen3.5 Omni Flash",
"description": "Omni-modal model for text, vision, audio, and multimodal agent tasks",
"family": "qwen3.5",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-03-30",
"last_updated": "2026-03-30",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 49152,
"input": 49152,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-omni-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-omni-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-35b-a3b:thinking": {
"id": "qwen3.5-35b-a3b:thinking",
"name": "Qwen3.5 35B A3B Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 81920
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 260096,
"input": 260096,
"output": 65536
},
"cost": {
"input": 0.225,
"output": 1.8,
"cache_read": 0.1125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-35b-a3b:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-35b-a3b:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-thinking:1024": {
"id": "claude-sonnet-4-thinking:1024",
"name": "Claude 4 Sonnet Thinking (1K)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-sonnet-4-thinking:1024\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-thinking:1024\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-1.5-vision-pro-32k": {
"id": "doubao-1.5-vision-pro-32k",
"name": "Doubao 1.5 Vision Pro 32k",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-11-20",
"last_updated": "2025-01-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"input": 32000,
"output": 8192
},
"cost": {
"input": 0.459,
"output": 1.377,
"cache_read": 0.2295
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/doubao-1.5-vision-pro-32k\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"doubao-1.5-vision-pro-32k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-0-lite-260215": {
"id": "doubao-seed-2-0-lite-260215",
"name": "Doubao Seed 2.0 Lite",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 32000
},
"cost": {
"input": 0.1462,
"output": 0.8738,
"cache_read": 0.0731
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/doubao-seed-2-0-lite-260215\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-lite-260215\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Gemma-4-26B-A4B-MeroMero:thinking": {
"id": "Gemma-4-26B-A4B-MeroMero:thinking",
"name": "Gemma 4 26B A4B MeroMero Thinking",
"description": "Gemma 4 26B A4B MeroMero with thinking enabled for more deliberate emotive dialogue, relationship scenes, creative writing, and multimodal roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Gemma-4-26B-A4B-MeroMero:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Gemma-4-26B-A4B-MeroMero:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-preview-04-17:thinking": {
"id": "gemini-2.5-flash-preview-04-17:thinking",
"name": "Gemini 2.5 Flash Preview Thinking",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"release_date": "2025-04-17",
"last_updated": "2025-04-17",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 3.5,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash-preview-04-17:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-preview-04-17:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5-20251001": {
"id": "claude-haiku-4-5-20251001",
"name": "Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-haiku-4-5-20251001\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5-20251001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claw-medium": {
"id": "claw-medium",
"name": "Claw Medium",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-11",
"last_updated": "2026-05-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claw-medium\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claw-medium\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4-long": {
"id": "glm-4-long",
"name": "GLM-4 Long",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-08-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 4096
},
"cost": {
"input": 0.2006,
"output": 0.2006,
"cache_read": 0.1003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/glm-4-long\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-4-long\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Gemma-4-31B-GarnetV2": {
"id": "Gemma-4-31B-GarnetV2",
"name": "Gemma 4 31B Garnet V2",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-05-01",
"last_updated": "2026-05-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.306,
"output": 0.306,
"cache_read": 0.153
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Gemma-4-31B-GarnetV2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Gemma-4-31B-GarnetV2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-preview-04-17": {
"id": "gemini-2.5-flash-preview-04-17",
"name": "Gemini 2.5 Flash Preview",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"release_date": "2025-04-17",
"last_updated": "2025-04-17",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash-preview-04-17\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-preview-04-17\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Gemma-4-31B-Claude-4.6-Opus-Reasoning-Distilled": {
"id": "Gemma-4-31B-Claude-4.6-Opus-Reasoning-Distilled",
"name": "Gemma 4 31B Claude 4.6 Opus Reasoning Distilled",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "claude",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-05-01",
"last_updated": "2026-05-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.306,
"output": 0.306,
"cache_read": 0.0306
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Gemma-4-31B-Claude-4.6-Opus-Reasoning-Distilled\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Gemma-4-31B-Claude-4.6-Opus-Reasoning-Distilled\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-27b:thinking": {
"id": "qwen3.8-27b:thinking",
"name": "Qwen3.8 27B Thinking",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.7,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.8-27b:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-27b:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-plus": {
"id": "qwen-plus",
"name": "Qwen Plus",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024,
"max": 81920
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01-25",
"last_updated": "2025-09-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 995904,
"input": 995904,
"output": 32768
},
"cost": {
"input": 0.3995,
"output": 1.2002,
"cache_read": 0.19975
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen-plus\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fastgpt": {
"id": "fastgpt",
"name": "Web Answer",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-23",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 7.5,
"output": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/fastgpt\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"fastgpt\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-nothinking": {
"id": "gemini-2.5-flash-nothinking",
"name": "Gemini 2.5 Flash (No Thinking)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-06-05",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash-nothinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-nothinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-122b-a10b": {
"id": "qwen3.5-122b-a10b",
"name": "Qwen3.5 122B A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 0.437,
"output": 3.496,
"cache_read": 0.103788
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-122b-a10b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-122b-a10b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-reasoner-cheaper": {
"id": "deepseek-reasoner-cheaper",
"name": "Deepseek R1 Cheaper",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-01-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 1.7,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-reasoner-cheaper\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-reasoner-cheaper\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar": {
"id": "sonar",
"name": "Perplexity Simple",
"description": "Fast web-grounded Sonar for current answers, citations, and lightweight retrieval",
"family": "sonar",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 127072,
"input": 127072,
"output": 114364
},
"cost": {
"input": 1,
"output": 1,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/sonar\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"sonar\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite-preview-09-2025-thinking": {
"id": "gemini-2.5-flash-lite-preview-09-2025-thinking",
"name": "Gemini 2.5 Flash Lite Preview (09/2025) – Thinking",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text",
"image",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash-lite-preview-09-2025-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite-preview-09-2025-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-thinking": {
"id": "claude-sonnet-4-thinking",
"name": "Claude 4 Sonnet Thinking",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-sonnet-4-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"auto-model-standard": {
"id": "auto-model-standard",
"name": "Auto model (Standard)",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"family": "auto",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2025-04-16",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 1000000
},
"cost": {
"input": 9.996,
"output": 19.992,
"cache_read": 4.998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/auto-model-standard\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"auto-model-standard\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-pro-image-preview": {
"id": "gemini-3-pro-image-preview",
"name": "Gemini 3 Pro Image",
"description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits",
"family": "gemini-pro",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-11-20",
"last_updated": "2025-11-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"input": 65536,
"output": 32768
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-3-pro-image-preview\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-pro-image-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar-reasoning-pro": {
"id": "sonar-reasoning-pro",
"name": "Perplexity Reasoning Pro",
"description": "Web-grounded Sonar for multi-step research questions that need cited reasoning",
"family": "sonar-reasoning",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 115200
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/sonar-reasoning-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"sonar-reasoning-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M1": {
"id": "MiniMax-M1",
"name": "MiniMax M1",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "minimax",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-01-08",
"last_updated": "2025-06-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 131072
},
"cost": {
"input": 0.1394,
"output": 1.3328,
"cache_read": 0.0697
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/MiniMax-M1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-30b-a3b-instruct-2507": {
"id": "qwen3-30b-a3b-instruct-2507",
"name": "Qwen3 30B A3B Instruct 2507",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-02-20",
"last_updated": "2025-02-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3-30b-a3b-instruct-2507\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-30b-a3b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Gemma-4-31B-DarkIdol": {
"id": "Gemma-4-31B-DarkIdol",
"name": "Gemma 4 31B DarkIdol",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-05-01",
"last_updated": "2026-05-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.306,
"output": 0.306,
"cache_read": 0.153
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Gemma-4-31B-DarkIdol\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Gemma-4-31B-DarkIdol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-plus:thinking": {
"id": "qwen3.7-plus:thinking",
"name": "Qwen3.7 Plus Thinking",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 983616,
"input": 983616,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.08,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.7-plus:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-plus:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"auto-model": {
"id": "auto-model",
"name": "Auto model",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"family": "auto",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2025-04-16",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 1000000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/auto-model\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"auto-model\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it-darkidol": {
"id": "gemma-4-31b-it-darkidol",
"name": "DarkIdol",
"description": "DarkIdol is a multimodal Gemma 4 31B creative finetune for expressive dialogue, long-form storytelling, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.45,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-31b-it-darkidol\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it-darkidol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1-thinking": {
"id": "claude-opus-4-1-thinking",
"name": "Claude 4.1 Opus Thinking",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-1-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-1-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.1v-thinking-flash": {
"id": "glm-4.1v-thinking-flash",
"name": "GLM 4.1V Thinking Flash",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "glm",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-09",
"last_updated": "2025-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 64000,
"input": 64000,
"output": 8192
},
"cost": {
"input": 0.3,
"output": 0.3,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/glm-4.1v-thinking-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.1v-thinking-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-4b": {
"id": "qwen3.5-4b",
"name": "Qwen3.5 4B",
"description": "Qwen3.5 4B is a compact open-weight multimodal model from Alibaba for reasoning, coding, visual understanding, tool use, and structured output.",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-16",
"last_updated": "2026-08-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-4b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1-thinking:32768": {
"id": "claude-opus-4-1-thinking:32768",
"name": "Claude 4.1 Opus Thinking (32K)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-1-thinking:32768\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-1-thinking:32768\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-31-24b-instruct": {
"id": "mistral-small-31-24b-instruct",
"name": "Mistral Small 31 24b Instruct",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "mistral-small",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-04-15",
"last_updated": "2025-04-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 102400
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistral-small-31-24b-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small-31-24b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it-fabled": {
"id": "gemma-4-31b-it-fabled",
"name": "Fabled",
"description": "Fabled is a multimodal Gemma 4 31B creative finetune for expressive dialogue, long-form storytelling, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.45,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-31b-it-fabled\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it-fabled\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"agnes-3.0-flash": {
"id": "agnes-3.0-flash",
"name": "Agnes 3.0 Flash",
"description": "Agnes 3.0 Flash is a low-cost model for coding, tool use, and multi-turn agent tasks. It supports text and image input, optional thinking, and a 512K-token context window.",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-09-09",
"last_updated": "2026-09-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 524288,
"input": 524288,
"output": 65536
},
"cost": {
"input": 0.05,
"output": 0.15,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/agnes-3.0-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"agnes-3.0-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-235b-a22b-instruct-original": {
"id": "qwen3-vl-235b-a22b-instruct-original",
"name": "Qwen3 VL 235B A22B Instruct Original",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 1.2,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3-vl-235b-a22b-instruct-original\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-235b-a22b-instruct-original\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-max-preview": {
"id": "qwen3.6-max-preview",
"name": "Qwen3.6 Max Preview",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-20",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 245760,
"input": 245760,
"output": 65536
},
"cost": {
"input": 1.04,
"output": 6.24,
"cache_read": 0.52
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.6-max-preview\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-preview-05-20": {
"id": "gemini-2.5-flash-preview-05-20",
"name": "Gemini 2.5 Flash 0520",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-05-20",
"last_updated": "2025-05-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048000,
"input": 1048000,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash-preview-05-20\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-preview-05-20\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mercury-2": {
"id": "mercury-2",
"name": "Mercury 2",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "mercury",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 50000
},
"cost": {
"input": 0.25,
"output": 0.75,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mercury-2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mercury-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it-gembrain": {
"id": "gemma-4-31b-it-gembrain",
"name": "Gembrain",
"description": "Gembrain is a multimodal Gemma 4 31B creative finetune for expressive dialogue, long-form storytelling, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.45,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-31b-it-gembrain\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it-gembrain\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-thinking:1024": {
"id": "claude-opus-4-thinking:1024",
"name": "Claude 4 Opus Thinking (1K)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-thinking:1024\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-thinking:1024\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max": {
"id": "qwen3.8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"input": 991000,
"output": 65536
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.8-max\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"longcat-2.0:thinking": {
"id": "longcat-2.0:thinking",
"name": "LongCat 2.0 Thinking",
"description": "Meituan LongCat-2.0, a reasoning model with tool calling and a 1M-token context window",
"family": "longcat",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048756,
"input": 1048756,
"output": 262144
},
"cost": {
"input": 0.75,
"output": 3,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/longcat-2.0:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"longcat-2.0:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-235b-a22b-thinking": {
"id": "qwen3-vl-235b-a22b-thinking",
"name": "Qwen3 VL 235B A22B Thinking",
"description": "Qwen vision-language thinking model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 81920
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3-vl-235b-a22b-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-235b-a22b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"celeris-1": {
"id": "celeris-1",
"name": "Celeris 1",
"description": "Celeris 1 is a diffusion language model built for ultra-low-latency classification, extraction, judging, query rewriting, and other short structured responses.",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-07-25",
"last_updated": "2026-07-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"input": 8192,
"output": 8192
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/celeris-1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"celeris-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-plus": {
"id": "qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991808,
"input": 991808,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.08,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.7-plus\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3-0324": {
"id": "deepseek-v3-0324",
"name": "DeepSeek Chat 0324",
"description": "March 2025 checkpoint of DeepSeek-V3 with improved reasoning and coding",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-03-24",
"last_updated": "2025-03-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 8192
},
"cost": {
"input": 0.2,
"output": 0.77,
"cache_read": 0.135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-v3-0324\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3-0324\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"brave-pro": {
"id": "brave-pro",
"name": "Brave (Pro)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-02-10",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"input": 8192,
"output": 8192
},
"cost": {
"input": 5,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/brave-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"brave-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it-novelist": {
"id": "gemma-4-31b-it-novelist",
"name": "Novelist",
"description": "Novelist is a multimodal Gemma 4 31B creative finetune for expressive dialogue, long-form storytelling, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.45,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-31b-it-novelist\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it-novelist\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-12b-it": {
"id": "gemma-4-12b-it",
"name": "Gemma 4 12B Instruct",
"description": "Google's Gemma 4 12B Instruct is an open-weight multimodal model for text, image, audio, and video understanding, with tool calling and structured output support.",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-01",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.25,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-12b-it\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-12b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar-pro": {
"id": "sonar-pro",
"name": "Perplexity Pro",
"description": "Deeper Sonar search model with broader retrieval and stronger synthesis",
"family": "sonar-pro",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 128000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/sonar-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"sonar-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-2b": {
"id": "qwen3.5-2b",
"name": "Qwen3.5 2B",
"description": "Qwen3.5 2B is a small open-weight multimodal model from Alibaba for efficient reasoning, coding, visual understanding, tool use, and JSON output.",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-16",
"last_updated": "2026-08-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.08,
"output": 0.16,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3.5-2b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-2b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-0-pro-260215": {
"id": "doubao-seed-2-0-pro-260215",
"name": "Doubao Seed 2.0 Pro",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 128000
},
"cost": {
"input": 0.782,
"output": 3.876,
"cache_read": 0.391
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/doubao-seed-2-0-pro-260215\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-pro-260215\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Gemma-4-31B-MeroMero-v2:thinking": {
"id": "Gemma-4-31B-MeroMero-v2:thinking",
"name": "Gemma 4 31B MeroMero v2 Thinking",
"description": "Gemma 4 31B MeroMero v2 with thinking enabled for more deliberate emotive dialogue, relationship scenes, creative writing, and multimodal roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-08-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.45,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Gemma-4-31B-MeroMero-v2:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Gemma-4-31B-MeroMero-v2:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar-deep-research": {
"id": "sonar-deep-research",
"name": "Perplexity Deep Research",
"description": "Sonar search model for autonomous research and citation-backed long-form reports",
"family": "sonar",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-02-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 115200
},
"cost": {
"input": 3.4,
"output": 13.6,
"cache_read": 1.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/sonar-deep-research\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"sonar-deep-research\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-instruct-fast": {
"id": "kimi-k2-instruct-fast",
"name": "Kimi K2 0711 Fast",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "kimi-k2",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-15",
"last_updated": "2025-07-15",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 16384
},
"cost": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/kimi-k2-instruct-fast\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-instruct-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-12b-it-semancer": {
"id": "gemma-4-12b-it-semancer",
"name": "Gemma 4 12B Semancer",
"description": "Gemma 4 12B Semancer is an open-weight roleplay finetune with image understanding, tool calling, optional reasoning, and a 131,072-token context window.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-09-09",
"last_updated": "2026-09-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.25,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-12b-it-semancer\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-12b-it-semancer\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Gemma-4-31B-MeroMero-v2": {
"id": "Gemma-4-31B-MeroMero-v2",
"name": "Gemma 4 31B MeroMero v2",
"description": "Gemma 4 31B MeroMero v2 is a LoRA finetune for emotive dialogue, relationship scenes, creative writing, and multimodal roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-08-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.45,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Gemma-4-31B-MeroMero-v2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Gemma-4-31B-MeroMero-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"auto-model-basic": {
"id": "auto-model-basic",
"name": "Auto model (Basic)",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"family": "auto",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2025-04-16",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 1000000
},
"cost": {
"input": 9.996,
"output": 19.992,
"cache_read": 4.998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/auto-model-basic\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"auto-model-basic\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-preview-05-20:thinking": {
"id": "gemini-2.5-flash-preview-05-20:thinking",
"name": "Gemini 2.5 Flash 0520 Thinking",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"release_date": "2025-05-20",
"last_updated": "2025-05-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048000,
"input": 1048000,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 3.5,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash-preview-05-20:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-preview-05-20:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro-exp-03-25": {
"id": "gemini-2.5-pro-exp-03-25",
"name": "Gemini 2.5 Pro Experimental 0325",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"release_date": "2025-03-25",
"last_updated": "2025-03-25",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-pro-exp-03-25\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro-exp-03-25\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-20250514": {
"id": "claude-sonnet-4-20250514",
"name": "Claude 4 Sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-sonnet-4-20250514\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-20250514\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-long": {
"id": "qwen-long",
"name": "Qwen Long 10M",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-08-01",
"last_updated": "2025-01-25",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 10000000,
"input": 10000000,
"output": 8192
},
"cost": {
"input": 0.1003,
"output": 0.408,
"cache_read": 0.05015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen-long\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen-long\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"phi-4-mini-instruct": {
"id": "phi-4-mini-instruct",
"name": "Phi 4 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "phi",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-26",
"last_updated": "2025-07-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 0.17,
"output": 0.68,
"cache_read": 0.085
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/phi-4-mini-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"phi-4-mini-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"hermes-low": {
"id": "hermes-low",
"name": "Hermes Low",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "hermes",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-11",
"last_updated": "2026-05-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/hermes-low\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"hermes-low\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nano-gpt-help": {
"id": "nano-gpt-help",
"name": "NanoGPT Help",
"description": "Text-only NanoGPT support assistant. Questions are processed by the Help inference provider; do not paste secrets or account credentials. Covers the website, models, API, pricing, memory, media generation, and support.",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-06-06",
"last_updated": "2026-06-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 6000,
"input": 6000,
"output": 512
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nano-gpt-help\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nano-gpt-help\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-12b-it-station-keeper": {
"id": "gemma-4-12b-it-station-keeper",
"name": "Gemma 4 12B StationKeeper",
"description": "Gemma 4 12B StationKeeper is an open-weight roleplay finetune with image understanding, tool calling, optional reasoning, and a 131,072-token context window.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-09-09",
"last_updated": "2026-09-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.25,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-12b-it-station-keeper\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-12b-it-station-keeper\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-max-2026-01-23": {
"id": "qwen3-max-2026-01-23",
"name": "Qwen3 Max 2026-01-23",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-01-26",
"last_updated": "2026-01-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 32768
},
"cost": {
"input": 1.2002,
"output": 6.001,
"cache_read": 0.6001
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen3-max-2026-01-23\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-max-2026-01-23\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro-preview-05-06": {
"id": "gemini-2.5-pro-preview-05-06",
"name": "Gemini 2.5 Pro Preview 0506",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"release_date": "2025-05-06",
"last_updated": "2025-05-06",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-2.5-pro-preview-05-06\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro-preview-05-06\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"brave": {
"id": "brave",
"name": "Brave (Answers)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-02-13",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"input": 8192,
"output": 8192
},
"cost": {
"input": 5,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/brave\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"brave\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-uncensored:thinking": {
"id": "gemma-4-26b-a4b-uncensored:thinking",
"name": "Gemma 4 26B A4B Uncensored Thinking",
"description": "Gemma 4 26B A4B Uncensored with thinking enabled for more deliberate coding, multimodal analysis, tool use, and long-context problem solving.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-26b-a4b-uncensored:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-uncensored:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"pokee-isaac": {
"id": "pokee-isaac",
"name": "Pokee-Isaac 28B",
"description": "Pokee-Isaac is a 28B agentic model with a roughly 10-million-token context window, function calling, and OpenAI-compatible structured output. Pokee bills in $0.01 increments, rounding each non-zero request up to the next cent.",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-04",
"last_updated": "2026-08-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 10000000,
"input": 10000000,
"output": 60000
},
"cost": {
"input": 0.15,
"output": 1,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/pokee-isaac\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"pokee-isaac\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1-thinking:8192": {
"id": "claude-opus-4-1-thinking:8192",
"name": "Claude 4.1 Opus Thinking (8K)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-1-thinking:8192\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-1-thinking:8192\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ernie-x1.1-preview": {
"id": "ernie-x1.1-preview",
"name": "ERNIE X1.1",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "ernie",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-10",
"last_updated": "2025-09-10",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 64000,
"input": 64000,
"output": 8192
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/ernie-x1.1-preview\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"ernie-x1.1-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-e2b-it": {
"id": "gemma-4-e2b-it",
"name": "Gemma 4 E2B Instruct",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 16384
},
"cost": {
"input": 0.02,
"output": 0.1,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-e2b-it\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-e2b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-exp-1206": {
"id": "gemini-exp-1206",
"name": "Gemini 2.0 Pro 1206",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gemini",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2097152,
"input": 2097152,
"output": 8192
},
"cost": {
"input": 1.258,
"output": 4.998,
"cache_read": 0.629
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemini-exp-1206\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-exp-1206\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it-isometry": {
"id": "gemma-4-31b-it-isometry",
"name": "Isometry",
"description": "Isometry is a multimodal Gemma 4 31B creative finetune for expressive dialogue, long-form storytelling, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.45,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-31b-it-isometry\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it-isometry\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-it-musica": {
"id": "gemma-4-26b-a4b-it-musica",
"name": "Musica",
"description": "Musica is a Gemma 4 26B A4B multimodal mixture-of-experts fine-tune for creative writing, expressive dialogue, and roleplay.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/gemma-4-26b-a4b-it-musica\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it-musica\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"hermes-medium": {
"id": "hermes-medium",
"name": "Hermes Medium",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "hermes",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-11",
"last_updated": "2026-05-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/hermes-medium\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"hermes-medium\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5-20251101": {
"id": "claude-opus-4-5-20251101",
"name": "Claude 4.5 Opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-01",
"last_updated": "2025-11-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/claude-opus-4-5-20251101\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-5-20251101\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepclaude": {
"id": "deepclaude",
"name": "DeepClaude",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-03-10",
"last_updated": "2025-02-01",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 8192
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepclaude\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepclaude\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qvq-max": {
"id": "qvq-max",
"name": "Qwen: QvQ Max",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-03-28",
"last_updated": "2025-03-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 8192
},
"cost": {
"input": 1.2,
"output": 4.8,
"cache_read": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qvq-max\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qvq-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"LatitudeGames/Wayfarer-Large-70B-Llama-3.3": {
"id": "LatitudeGames/Wayfarer-Large-70B-Llama-3.3",
"name": "Llama 3.3 70B Wayfarer",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-02-20",
"last_updated": "2025-02-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 0.7,
"output": 0.7,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/LatitudeGames/Wayfarer-Large-70B-Llama-3.3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"LatitudeGames/Wayfarer-Large-70B-Llama-3.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-397b-a17b-thinking": {
"id": "qwen/qwen3.5-397b-a17b-thinking",
"name": "Qwen3.5 397B A17B Thinking",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 81920
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 258048,
"input": 258048,
"output": 65536
},
"cost": {
"input": 0.6,
"output": 3.6,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.5-397b-a17b-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-397b-a17b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-plus": {
"id": "qwen/qwen3-coder-plus",
"name": "Qwen3 Coder Plus",
"description": "Hosted Qwen coder for software agents, repo edits, and long-context code",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3-coder-plus\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-next-80b-a3b-thinking": {
"id": "qwen/qwen3-next-80b-a3b-thinking",
"name": "Qwen3 Next 80B A3B (Thinking)",
"description": "Efficient Qwen thinking model for local reasoning, math, and coding agents",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.65,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3-next-80b-a3b-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-next-80b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-9b": {
"id": "qwen/qwen3.5-9b",
"name": "Qwen3.5 9B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024,
"max": 81920
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 65536
},
"cost": {
"input": 0.05,
"output": 0.15,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.5-9b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-flash": {
"id": "qwen/qwen3-coder-flash",
"name": "Qwen3 Coder Flash",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3-coder-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-14b": {
"id": "qwen/qwen3-14b",
"name": "Qwen 3 14b",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 41000,
"input": 41000,
"output": 32768
},
"cost": {
"input": 0.08,
"output": 0.24,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3-14b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-14b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/Qwen3-8B": {
"id": "qwen/Qwen3-8B",
"name": "Qwen 3 8B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 41000,
"input": 41000,
"output": 32768
},
"cost": {
"input": 0.47,
"output": 0.47,
"cache_read": 0.235
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/Qwen3-8B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/Qwen3-8B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-27b-obliterated": {
"id": "qwen/qwen3.8-27b-obliterated",
"name": "Qwen 3.8 27B Obliterated",
"description": "Qwen 3.8 27B Obliterated is an open-weight multimodal model LoRA-tuned for fewer refusals across chat, coding, reasoning, tool use, and long-context work.",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-29",
"last_updated": "2026-08-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.8-27b-obliterated\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-27b-obliterated\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-27b-queen": {
"id": "qwen/qwen3.8-27b-queen",
"name": "Qwen 3.8 27B Queen",
"description": "Qwen 3.8 27B Queen is an open-weight roleplay finetune with image understanding, tool calling, optional reasoning, and a 262,144-token context window.",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-09-09",
"last_updated": "2026-09-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.8-27b-queen\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-27b-queen\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-plus-thinking": {
"id": "qwen/qwen3.5-plus-thinking",
"name": "Qwen3.5 Plus Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 81920
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 983616,
"input": 983616,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 2.4,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.5-plus-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-plus-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-32b": {
"id": "qwen/qwen3-32b",
"name": "Qwen 3 32b",
"description": "Dense open Qwen model for self-hosted chat, reasoning, and coding",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 41000,
"input": 41000,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3-32b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder": {
"id": "qwen/qwen3-coder",
"name": "Qwen 3 Coder 480B",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"input": 262000,
"output": 65536
},
"cost": {
"input": 0.13,
"output": 0.5,
"cache_read": 0.065
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3-coder\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-next": {
"id": "qwen/qwen3-coder-next",
"name": "Qwen3 Coder Next",
"description": "Open-weight Qwen coding model for agents, repository edits, and multi-turn tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2026-02-03",
"last_updated": "2026-02-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 0.2,
"output": 1.5,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3-coder-next\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-397b-a17b": {
"id": "qwen/qwen3.5-397b-a17b",
"name": "Qwen3.5 397B A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 258048,
"input": 258048,
"output": 65536
},
"cost": {
"input": 0.6,
"output": 3.6,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/Qwen3-235B-A22B-Instruct-2507": {
"id": "qwen/Qwen3-235B-A22B-Instruct-2507",
"name": "Qwen 3 235b A22B 2507",
"description": "Updated large open Qwen3 MoE instruct model for multilingual chat, coding, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-21",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 235929
},
"cost": {
"input": 0.13,
"output": 0.5,
"cache_read": 0.065
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/Qwen3-235B-A22B-Instruct-2507\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/Qwen3-235B-A22B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/Qwen3-VL-235B-A22B-Instruct": {
"id": "qwen/Qwen3-VL-235B-A22B-Instruct",
"name": "Qwen3 VL 235B A22B Instruct",
"description": "Qwen vision-language instruct model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/Qwen3-VL-235B-A22B-Instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/Qwen3-VL-235B-A22B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-27b-fable": {
"id": "qwen/qwen3.8-27b-fable",
"name": "Qwen 3.8 27B Fable",
"description": "Qwen 3.8 27B Fable is an open-weight multimodal creative finetune for expressive dialogue, long-form storytelling, character work, and roleplay.",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-29",
"last_updated": "2026-08-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.8-27b-fable\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-27b-fable\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-max": {
"id": "qwen/qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 32768
},
"cost": {
"input": 1.2002,
"output": 6.001,
"cache_read": 0.6001
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3-max\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/Qwen3-Next-80B-A3B-Instruct": {
"id": "qwen/Qwen3-Next-80B-A3B-Instruct",
"name": "Qwen3 Next 80B A3B (Instruct)",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 235929
},
"cost": {
"input": 0.15,
"output": 0.65,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/Qwen3-Next-80B-A3B-Instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/Qwen3-Next-80B-A3B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-2.4t-a95b": {
"id": "qwen/qwen3.8-2.4t-a95b",
"name": "Qwen3.8 2.4T A95B (Max)",
"description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 991000,
"input": 991000,
"output": 65536
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.8-2.4t-a95b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-2.4t-a95b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-27b-uncensored": {
"id": "qwen/qwen3.8-27b-uncensored",
"name": "Qwen 3.8 27B Uncensored",
"description": "Qwen 3.8 27B Uncensored is an NVFP4 open-weight multimodal model LoRA-tuned for fewer refusals across chat, coding, reasoning, tool use, and long-context work.",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-29",
"last_updated": "2026-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.8-27b-uncensored\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-27b-uncensored\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-30b-a3b": {
"id": "qwen/qwen3-30b-a3b",
"name": "Qwen3 30B A3B",
"description": "Sparse MoE Qwen model with 3B active parameters for efficient chat and reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-04-28",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 41000,
"input": 41000,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3-30b-a3b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/Qwen3-235B-A22B-Thinking-2507": {
"id": "qwen/Qwen3-235B-A22B-Thinking-2507",
"name": "Qwen 3 235b A22B 2507 Thinking",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2025-09-11",
"last_updated": "2025-09-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 117964
},
"cost": {
"input": 0.3,
"output": 0.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/Qwen3-235B-A22B-Thinking-2507\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/Qwen3-235B-A22B-Thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-2.5-72b-instruct": {
"id": "qwen/qwen-2.5-72b-instruct",
"name": "Qwen2.5 72B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2025-07-03",
"last_updated": "2025-07-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 8192
},
"cost": {
"input": 0.357,
"output": 0.408,
"cache_read": 0.1785
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen-2.5-72b-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-2.5-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b": {
"id": "qwen/qwen3-235b-a22b",
"name": "Qwen 3 235b A22B",
"description": "Large open Qwen MoE for multilingual reasoning, coding, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.3,
"output": 0.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3-235b-a22b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/Qwen3.6-35B-A3B": {
"id": "qwen/Qwen3.6-35B-A3B",
"name": "Qwen3.6 35B A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.112,
"output": 0.8,
"cache_read": 0.056
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/Qwen3.6-35B-A3B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/Qwen3.6-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/Qwen3.6-35B-A3B:thinking": {
"id": "qwen/Qwen3.6-35B-A3B:thinking",
"name": "Qwen3.6 35B A3B Thinking",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.112,
"output": 0.8,
"cache_read": 0.056
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/Qwen3.6-35B-A3B:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/Qwen3.6-35B-A3B:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-27b-uncensored:thinking": {
"id": "qwen/qwen3.8-27b-uncensored:thinking",
"name": "Qwen 3.8 27B Uncensored Thinking",
"description": "Qwen 3.8 27B Uncensored with thinking enabled for more deliberate creative work, coding, multimodal analysis, tool use, and long-context problem solving.",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-29",
"last_updated": "2026-08-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.8-27b-uncensored:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-27b-uncensored:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-plus": {
"id": "qwen/qwen3.5-plus",
"name": "Qwen3.5 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 983616,
"input": 983616,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 2.4,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.5-plus\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-27b-obliterated:thinking": {
"id": "qwen/qwen3.8-27b-obliterated:thinking",
"name": "Qwen 3.8 27B Obliterated Thinking",
"description": "Qwen 3.8 27B Obliterated with thinking enabled for more deliberate creative work, coding, multimodal analysis, tool use, and long-context problem solving.",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-29",
"last_updated": "2026-08-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/qwen3.8-27b-obliterated:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-27b-obliterated:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/Qwen2.5-Coder-32B-Instruct": {
"id": "qwen/Qwen2.5-Coder-32B-Instruct",
"name": "Qwen 2.5 Coder 32b",
"description": "Open coding-focused Qwen model for code generation, repair, and repository reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2024-11-12",
"last_updated": "2024-11-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"input": 32000,
"output": 8192
},
"cost": {
"input": 0.2006,
"output": 0.2006,
"cache_read": 0.1003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/qwen/Qwen2.5-Coder-32B-Instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"qwen/Qwen2.5-Coder-32B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MarinaraSpaghetti/NemoMix-Unleashed-12B": {
"id": "MarinaraSpaghetti/NemoMix-Unleashed-12B",
"name": "NemoMix 12B Unleashed",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-nemo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 8192
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/MarinaraSpaghetti/NemoMix-Unleashed-12B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"MarinaraSpaghetti/NemoMix-Unleashed-12B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aion-labs/aion-2.0": {
"id": "aion-labs/aion-2.0",
"name": "AionLabs: Aion-2.0",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 0.8,
"output": 1.6,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/aion-labs/aion-2.0\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"aion-labs/aion-2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aion-labs/aion-rp-llama-3.1-8b": {
"id": "aion-labs/aion-rp-llama-3.1-8b",
"name": "Llama 3.1 8b (uncensored)",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 0.8,
"output": 1.6,
"cache_read": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/aion-labs/aion-rp-llama-3.1-8b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"aion-labs/aion-rp-llama-3.1-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aion-labs/aion-3.0": {
"id": "aion-labs/aion-3.0",
"name": "AionLabs: Aion 3.0",
"description": "Aion 3.0 is a GLM-family collaborative generation model tuned for immersive roleplay and storytelling, with stronger narrative structure, tension, conflict, and nuanced mature themes.",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-07",
"last_updated": "2026-07-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 3,
"output": 6,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/aion-labs/aion-3.0\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"aion-labs/aion-3.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aion-labs/aion-3.0-mini": {
"id": "aion-labs/aion-3.0-mini",
"name": "AionLabs: Aion 3.0 Mini",
"description": "Aion 3.0 Mini is a DeepSeek-family collaborative generation model tuned for immersive roleplay and storytelling, with stronger narrative structure, tension, conflict, and nuanced mature themes.",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-07",
"last_updated": "2026-07-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 0.7,
"output": 1.4,
"cache_read": 0.18
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/aion-labs/aion-3.0-mini\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"aion-labs/aion-3.0-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nothingiisreal/L3.1-70B-Celeste-V0.1-BF16": {
"id": "nothingiisreal/L3.1-70B-Celeste-V0.1-BF16",
"name": "Llama 3.1 70B Celeste v0.1",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nothingiisreal/L3.1-70B-Celeste-V0.1-BF16\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nothingiisreal/L3.1-70B-Celeste-V0.1-BF16\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ornith-ai/ornith-1.5-35b-a3b": {
"id": "ornith-ai/ornith-1.5-35b-a3b",
"name": "Ornith 1.5 35B",
"description": "Ornith 1.5 35B A3B is an open-weight mixture-of-experts model for agentic coding, tool use, image understanding, and long-context work. This variant disables thinking for faster direct responses.",
"family": "ornith",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-29",
"last_updated": "2026-08-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/ornith-ai/ornith-1.5-35b-a3b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"ornith-ai/ornith-1.5-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ornith-ai/ornith-1.5-35b-a3b:thinking": {
"id": "ornith-ai/ornith-1.5-35b-a3b:thinking",
"name": "Ornith 1.5 35B Thinking",
"description": "Ornith 1.5 35B A3B is an open-weight mixture-of-experts model for agentic coding, reasoning, tool use, image understanding, and long-context work. This variant enables thinking by default.",
"family": "ornith",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-29",
"last_updated": "2026-08-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/ornith-ai/ornith-1.5-35b-a3b:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"ornith-ai/ornith-1.5-35b-a3b:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"abliteration-ai/abliterated-model-large": {
"id": "abliteration-ai/abliterated-model-large",
"name": "Abliterated Model Large",
"description": "Abliteration.ai's large text reasoning model is derived from GLM-5.2 and supports native tool calling, structured output, automatic prompt caching, and a one-million-token context window.",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-27",
"last_updated": "2026-08-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 999990
},
"cost": {
"input": 5,
"output": 5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/abliteration-ai/abliterated-model-large\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"abliteration-ai/abliterated-model-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"abliteration-ai/abliterated-model-large-v2": {
"id": "abliteration-ai/abliterated-model-large-v2",
"name": "Abliterated Model Large V2",
"description": "Abliteration.ai's default large text reasoning model is derived from GLM-5.3 for harder reasoning and evaluation workloads, with automatic prompt caching and a one-million-token context window.",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-31",
"last_updated": "2026-08-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 999990
},
"cost": {
"input": 5,
"output": 5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/abliteration-ai/abliterated-model-large-v2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"abliteration-ai/abliterated-model-large-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"abliteration-ai/abliterated-model": {
"id": "abliteration-ai/abliterated-model",
"name": "Abliterated Model",
"description": "Abliteration.ai's multimodal reasoning model supports text and image input, structured output, automatic prompt caching, and a 262K-token context window.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": false,
"structured_output": true,
"release_date": "2026-08-27",
"last_updated": "2026-08-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"input": 262144,
"output": 262134
},
"cost": {
"input": 3,
"output": 3,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/abliteration-ai/abliterated-model\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"abliteration-ai/abliterated-model\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"undi95/remm-slerp-l2-13b": {
"id": "undi95/remm-slerp-l2-13b",
"name": "ReMM SLERP 13B",
"description": "Open Llama multimodal model for image understanding and text reasoning",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 6144,
"input": 6144,
"output": 4096
},
"cost": {
"input": 0.799,
"output": 1.207,
"cache_read": 0.3995
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/undi95/remm-slerp-l2-13b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"undi95/remm-slerp-l2-13b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"chutesai/Mistral-Small-3.2-24B-Instruct-2506": {
"id": "chutesai/Mistral-Small-3.2-24B-Instruct-2506",
"name": "Mistral Small 3.2 24b Instruct",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "chutesai",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-04-15",
"last_updated": "2025-04-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 0.2,
"output": 0.4,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/chutesai/Mistral-Small-3.2-24B-Instruct-2506\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"chutesai/Mistral-Small-3.2-24B-Instruct-2506\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun-ai/step-3.5-flash": {
"id": "stepfun-ai/step-3.5-flash",
"name": "Step 3.5 Flash",
"description": "StepFun flash lane for quick multimodal reasoning and coding assistance",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-29",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/stepfun-ai/step-3.5-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"stepfun-ai/step-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun-ai/step-3.5-flash-2603": {
"id": "stepfun-ai/step-3.5-flash-2603",
"name": "Step 3.5 Flash 2603",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/stepfun-ai/step-3.5-flash-2603\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"stepfun-ai/step-3.5-flash-2603\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"pamanseau/OpenReasoning-Nemotron-32B": {
"id": "pamanseau/OpenReasoning-Nemotron-32B",
"name": "OpenReasoning Nemotron 32B",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/pamanseau/OpenReasoning-Nemotron-32B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"pamanseau/OpenReasoning-Nemotron-32B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.1:thinking": {
"id": "deepseek-ai/DeepSeek-V3.1:thinking",
"name": "DeepSeek V3.1 Thinking",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 0.2,
"output": 0.7,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-ai/DeepSeek-V3.1:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.1:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/deepseek-v3.2-exp-thinking": {
"id": "deepseek-ai/deepseek-v3.2-exp-thinking",
"name": "DeepSeek V3.2 Exp Thinking",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"input": 163840,
"output": 65536
},
"cost": {
"input": 0.28,
"output": 0.42,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-ai/deepseek-v3.2-exp-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/deepseek-v3.2-exp-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.1": {
"id": "deepseek-ai/DeepSeek-V3.1",
"name": "DeepSeek V3.1",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2025-07-26",
"last_updated": "2025-07-26",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 0.2,
"output": 0.7,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-ai/DeepSeek-V3.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.1-Terminus:thinking": {
"id": "deepseek-ai/DeepSeek-V3.1-Terminus:thinking",
"name": "DeepSeek V3.1 Terminus (Thinking)",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2024-01-01",
"last_updated": "2025-09-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 0.7,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-ai/DeepSeek-V3.1-Terminus:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.1-Terminus:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-R1-0528": {
"id": "deepseek-ai/DeepSeek-R1-0528",
"name": "DeepSeek R1 0528",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2025-05-28",
"last_updated": "2025-05-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"input": 163840,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.7,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-ai/DeepSeek-R1-0528\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-R1-0528\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/deepseek-v3.2-exp": {
"id": "deepseek-ai/deepseek-v3.2-exp",
"name": "DeepSeek V3.2 Exp",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"input": 163840,
"output": 65536
},
"cost": {
"input": 0.28,
"output": 0.42,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-ai/deepseek-v3.2-exp\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/deepseek-v3.2-exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.1-Terminus": {
"id": "deepseek-ai/DeepSeek-V3.1-Terminus",
"name": "DeepSeek V3.1 Terminus",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2025-08-02",
"last_updated": "2025-08-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 0.7,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek-ai/DeepSeek-V3.1-Terminus\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.1-Terminus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"VongolaChouko/Starcannon-Unleashed-12B-v1.0": {
"id": "VongolaChouko/Starcannon-Unleashed-12B-v1.0",
"name": "Mistral Nemo Starcannon 12b v1",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-nemo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/VongolaChouko/Starcannon-Unleashed-12B-v1.0\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"VongolaChouko/Starcannon-Unleashed-12B-v1.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-s-2.1:thinking": {
"id": "poolside/laguna-s-2.1:thinking",
"name": "Laguna S 2.1 Thinking",
"description": "Agentic coding model from Poolside in the XS size class for local deployment",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/poolside/laguna-s-2.1:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-s-2.1:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-s-2.1": {
"id": "poolside/laguna-s-2.1",
"name": "Laguna S 2.1",
"description": "Agentic coding model from Poolside in the XS size class for local deployment",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/poolside/laguna-s-2.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-s-2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"featherless-ai/Qwerky-72B": {
"id": "featherless-ai/Qwerky-72B",
"name": "Qwerky 72B",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"family": "qwerky",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-03-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"input": 32000,
"output": 8192
},
"cost": {
"input": 0.5,
"output": 0.5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/featherless-ai/Qwerky-72B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"featherless-ai/Qwerky-72B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun/step-3.7-flash:thinking": {
"id": "stepfun/step-3.7-flash:thinking",
"name": "Step 3.7 Flash Thinking",
"description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 256000
},
"cost": {
"input": 0.2,
"output": 1.15,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/stepfun/step-3.7-flash:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"stepfun/step-3.7-flash:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mlabonne/NeuralDaredevil-8B-abliterated": {
"id": "mlabonne/NeuralDaredevil-8B-abliterated",
"name": "Neural Daredevil 8B abliterated",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"input": 8192,
"output": 8192
},
"cost": {
"input": 0.44,
"output": 0.44,
"cache_read": 0.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mlabonne/NeuralDaredevil-8B-abliterated\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mlabonne/NeuralDaredevil-8B-abliterated\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/ministral-14b-2512": {
"id": "mistralai/ministral-14b-2512",
"name": "Ministral 14B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-04",
"last_updated": "2025-12-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/ministral-14b-2512\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/ministral-14b-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-large": {
"id": "mistralai/mistral-large",
"name": "Mistral Large 2411",
"description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
"family": "mistral-large",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-02-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 102400
},
"cost": {
"input": 2.006,
"output": 6.001,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/mistral-large\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/codestral-2508": {
"id": "mistralai/codestral-2508",
"name": "Codestral 2508",
"description": "Mistral coding model for code completion, generation, and developer workflows",
"family": "codestral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-08-01",
"last_updated": "2025-08-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 0.9,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/codestral-2508\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/codestral-2508\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-large-3-675b-instruct-2512": {
"id": "mistralai/mistral-large-3-675b-instruct-2512",
"name": "Mistral Large 3 675B",
"description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
"family": "mistral-large",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-25",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 256000
},
"cost": {
"input": 1,
"output": 3,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/mistral-large-3-675b-instruct-2512\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-large-3-675b-instruct-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Devstral-Small-2505": {
"id": "mistralai/Devstral-Small-2505",
"name": "Mistral Devstral Small 2505",
"description": "Mistral coding agent model for repository tasks and software engineering workflows",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-08-02",
"last_updated": "2025-08-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 8192
},
"cost": {
"input": 0.06,
"output": 0.06,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/Devstral-Small-2505\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Devstral-Small-2505\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/devstral-2-123b-instruct-2512": {
"id": "mistralai/devstral-2-123b-instruct-2512",
"name": "Devstral 2 123B",
"description": "Mistral coding agent model for repository tasks and software engineering workflows",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-09",
"last_updated": "2025-12-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 1.4,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/devstral-2-123b-instruct-2512\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/devstral-2-123b-instruct-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-saba": {
"id": "mistralai/mistral-saba",
"name": "Mistral Saba",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-02-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 26214
},
"cost": {
"input": 0.1989,
"output": 0.595,
"cache_read": 0.09945
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/mistral-saba\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-saba\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/ministral-14b-instruct-2512": {
"id": "mistralai/ministral-14b-instruct-2512",
"name": "Ministral 3 14B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/ministral-14b-instruct-2512\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/ministral-14b-instruct-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-small-4-119b-2603:thinking": {
"id": "mistralai/mistral-small-4-119b-2603:thinking",
"name": "Mistral Small 4 119B Thinking",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.4,
"output": 1.4,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/mistral-small-4-119b-2603:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-small-4-119b-2603:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-small-4-119b-2603": {
"id": "mistralai/mistral-small-4-119b-2603",
"name": "Mistral Small 4 119B",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.4,
"output": 1.4,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/mistral-small-4-119b-2603\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-small-4-119b-2603\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/ministral-3b-2512": {
"id": "mistralai/ministral-3b-2512",
"name": "Ministral 3B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-04",
"last_updated": "2025-12-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.1,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/ministral-3b-2512\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/ministral-3b-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-medium-3": {
"id": "mistralai/mistral-medium-3",
"name": "Mistral Medium 3",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-medium",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/mistral-medium-3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-medium-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-small-24b-instruct-2501": {
"id": "mistralai/mistral-small-24b-instruct-2501",
"name": "Mistral Small 24B",
"description": "Mistral Small 24B hosted by IONOS in Berlin, Germany. Zero data retention.",
"family": "mistral-small",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 8192
},
"cost": {
"input": 0.1155,
"output": 0.3465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/mistral-small-24b-instruct-2501\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-small-24b-instruct-2501\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mixtral-8x22b-instruct-v0.1": {
"id": "mistralai/mixtral-8x22b-instruct-v0.1",
"name": "Mixtral 8x22B",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mixtral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"input": 65536,
"output": 52428
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/mixtral-8x22b-instruct-v0.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mixtral-8x22b-instruct-v0.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Mistral-Nemo-Instruct-2407": {
"id": "mistralai/Mistral-Nemo-Instruct-2407",
"name": "Mistral Nemo",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-nemo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.1003,
"output": 0.1207,
"cache_read": 0.05015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/Mistral-Nemo-Instruct-2407\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Mistral-Nemo-Instruct-2407\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/ministral-8b-2512": {
"id": "mistralai/ministral-8b-2512",
"name": "Ministral 8B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-04",
"last_updated": "2025-12-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/ministral-8b-2512\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/ministral-8b-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-medium-3.1": {
"id": "mistralai/mistral-medium-3.1",
"name": "Mistral Medium 3.1",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-medium",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-05",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistralai/mistral-medium-3.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-medium-3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ReadyArt/MS3.2-The-Omega-Directive-24B-Unslop-v2.0": {
"id": "ReadyArt/MS3.2-The-Omega-Directive-24B-Unslop-v2.0",
"name": "Omega Directive 24B Unslop v2.0",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 0.5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/ReadyArt/MS3.2-The-Omega-Directive-24B-Unslop-v2.0\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"ReadyArt/MS3.2-The-Omega-Directive-24B-Unslop-v2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated": {
"id": "huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated",
"name": "DeepSeek R1 Llama 70B Abliterated",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-01-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.7,
"output": 0.7,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"huihui-ai/DeepSeek-R1-Distill-Qwen-32B-abliterated": {
"id": "huihui-ai/DeepSeek-R1-Distill-Qwen-32B-abliterated",
"name": "DeepSeek R1 Qwen Abliterated",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-01-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 1.4,
"output": 1.4,
"cache_read": 0.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/huihui-ai/DeepSeek-R1-Distill-Qwen-32B-abliterated\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"huihui-ai/DeepSeek-R1-Distill-Qwen-32B-abliterated\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"huihui-ai/Llama-3.3-70B-Instruct-abliterated": {
"id": "huihui-ai/Llama-3.3-70B-Instruct-abliterated",
"name": "Llama 3.3 70B Instruct abliterated",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-08-08",
"last_updated": "2025-08-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 0.7,
"output": 0.7,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/huihui-ai/Llama-3.3-70B-Instruct-abliterated\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"huihui-ai/Llama-3.3-70B-Instruct-abliterated\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"huihui-ai/Qwen2.5-32B-Instruct-abliterated": {
"id": "huihui-ai/Qwen2.5-32B-Instruct-abliterated",
"name": "Qwen 2.5 32B Abliterated",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-01-06",
"last_updated": "2025-01-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 8192
},
"cost": {
"input": 0.7,
"output": 0.7,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/huihui-ai/Qwen2.5-32B-Instruct-abliterated\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"huihui-ai/Qwen2.5-32B-Instruct-abliterated\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5-pro-crof": {
"id": "xiaomi/mimo-v2.5-pro-crof",
"name": "MiMo V2.5 Pro (Crof)",
"description": "MiMo V2.5 Pro is Xiaomi's long-context flagship general model for coding and agentic orchestration. This separately served variant is intended for users concerned about censorship on the regular Xiaomi MiMo V2.5 Pro, and it is included in the NanoGPT subscription.",
"family": "mimo-v2.5-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-23",
"last_updated": "2026-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 131072
},
"cost": {
"input": 0.4,
"output": 0.8,
"cache_read": 0.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/xiaomi/mimo-v2.5-pro-crof\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5-pro-crof\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5:thinking": {
"id": "xiaomi/mimo-v2.5:thinking",
"name": "MiMo V2.5 Thinking",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/xiaomi/mimo-v2.5:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5-pro-crof:thinking": {
"id": "xiaomi/mimo-v2.5-pro-crof:thinking",
"name": "MiMo V2.5 Pro Thinking (Crof)",
"description": "MiMo V2.5 Pro with Xiaomi thinking enabled for coding, long-context reasoning, and agentic orchestration. This separately served thinking variant is intended for users concerned about censorship on the regular Xiaomi MiMo V2.5 Pro, and it is included in the NanoGPT subscription.",
"family": "mimo-v2.5-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-23",
"last_updated": "2026-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 131072
},
"cost": {
"input": 0.4,
"output": 0.8,
"cache_read": 0.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/xiaomi/mimo-v2.5-pro-crof:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5-pro-crof:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5-pro:thinking": {
"id": "xiaomi/mimo-v2.5-pro:thinking",
"name": "MiMo V2.5 Pro Thinking",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.0036,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/xiaomi/mimo-v2.5-pro:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5-pro:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5": {
"id": "xiaomi/mimo-v2.5",
"name": "MiMo V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/xiaomi/mimo-v2.5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5-pro": {
"id": "xiaomi/mimo-v2.5-pro",
"name": "MiMo V2.5 Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.0036,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/xiaomi/mimo-v2.5-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"shisa-ai/shisa-v2-llama3.3-70b": {
"id": "shisa-ai/shisa-v2-llama3.3-70b",
"name": "Shisa V2 Llama 3.3 70B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-26",
"last_updated": "2025-07-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 0.5,
"output": 0.5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/shisa-ai/shisa-v2-llama3.3-70b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"shisa-ai/shisa-v2-llama3.3-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"shisa-ai/shisa-v2.1-llama3.3-70b": {
"id": "shisa-ai/shisa-v2.1-llama3.3-70b",
"name": "Shisa V2.1 Llama 3.3 70B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 0.5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/shisa-ai/shisa-v2.1-llama3.3-70b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"shisa-ai/shisa-v2.1-llama3.3-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m3:thinking": {
"id": "minimax/minimax-m3:thinking",
"name": "MiniMax M3 Thinking",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512000,
"input": 512000,
"output": 80000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/minimax/minimax-m3:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m3:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.1": {
"id": "minimax/minimax-m2.1",
"name": "MiniMax M2.1",
"description": "Earlier MiniMax agent model for practical coding and productivity tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 131072
},
"cost": {
"input": 0.33,
"output": 1.32,
"cache_read": 0.165
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/minimax/minimax-m2.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7": {
"id": "minimax/minimax-m2.7",
"name": "MiniMax M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"input": 204800,
"output": 131072
},
"cost": {
"input": 0.315,
"output": 1.26,
"cache_read": 0.1575
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/minimax/minimax-m2.7\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7-turbo": {
"id": "minimax/minimax-m2.7-turbo",
"name": "MiniMax M2.7 Turbo",
"description": "Efficient MiniMax model for quick assistance, coding, and routine automation",
"family": "minimax-m2.7",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"input": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/minimax/minimax-m2.7-turbo\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5": {
"id": "minimax/minimax-m2.5",
"name": "MiniMax M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"input": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/minimax/minimax-m2.5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m3": {
"id": "minimax/minimax-m3",
"name": "MiniMax M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512000,
"input": 512000,
"output": 80000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/minimax/minimax-m3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2-her": {
"id": "minimax/minimax-m2-her",
"name": "MiniMax M2-her",
"description": "MiniMax M2 variant tuned for conversational and character-driven agent interactions",
"family": "minimax",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-01-23",
"last_updated": "2026-01-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 65532,
"input": 65532,
"output": 2048
},
"cost": {
"input": 0.302,
"output": 1.207,
"cache_read": 0.151
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/minimax/minimax-m2-her\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2-her\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-01": {
"id": "minimax/minimax-01",
"name": "MiniMax 01",
"description": "MiniMax multimodal coding model for long-context reasoning and agent tasks",
"family": "minimax",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-01-15",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000192,
"input": 1000192,
"output": 16384
},
"cost": {
"input": 0.1394,
"output": 1.122,
"cache_read": 0.0697
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/minimax/minimax-01\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-01\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-latest": {
"id": "minimax/minimax-latest",
"name": "MiniMax Latest",
"description": "MiniMax multimodal coding model for long-context reasoning and agent tasks",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-03",
"last_updated": "2026-05-03",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512000,
"input": 512000,
"output": 80000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/minimax/minimax-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Sao10K/L3-8B-Stheno-v3.2": {
"id": "Sao10K/L3-8B-Stheno-v3.2",
"name": "Sao10K Stheno 8b",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-11-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.2006,
"output": 0.2006,
"cache_read": 0.1003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Sao10K/L3-8B-Stheno-v3.2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Sao10K/L3-8B-Stheno-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Sao10K/L3.3-70B-Euryale-v2.3": {
"id": "Sao10K/L3.3-70B-Euryale-v2.3",
"name": "Llama 3.3 70B Euryale",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 20480,
"input": 20480,
"output": 16384
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Sao10K/L3.3-70B-Euryale-v2.3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Sao10K/L3.3-70B-Euryale-v2.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Sao10K/L3.1-70B-Euryale-v2.2": {
"id": "Sao10K/L3.1-70B-Euryale-v2.2",
"name": "Llama 3.1 70B Euryale",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 20480,
"input": 20480,
"output": 16384
},
"cost": {
"input": 0.306,
"output": 0.357,
"cache_read": 0.153
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Sao10K/L3.1-70B-Euryale-v2.2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Sao10K/L3.1-70B-Euryale-v2.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Sao10K/L3.1-70B-Hanami-x1": {
"id": "Sao10K/L3.1-70B-Hanami-x1",
"name": "Llama 3.1 70B Hanami",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Sao10K/L3.1-70B-Hanami-x1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Sao10K/L3.1-70B-Hanami-x1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.6-27b:thinking": {
"id": "alibaba/qwen3.6-27b:thinking",
"name": "Qwen3.6 27B Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 131072
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 260096,
"input": 260096,
"output": 65536
},
"cost": {
"input": 0.203,
"output": 2.24,
"cache_read": 0.1015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/alibaba/qwen3.6-27b:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.6-27b:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.6-27b": {
"id": "alibaba/qwen3.6-27b",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 260096,
"input": 260096,
"output": 65536
},
"cost": {
"input": 0.203,
"output": 2.24,
"cache_read": 0.1015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/alibaba/qwen3.6-27b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.8-max-0902": {
"id": "alibaba/qwen3.8-max-0902",
"name": "Qwen3.8 Max 0902",
"description": "2026-09-02 upgraded snapshot of Qwen3.8 Max with stronger coding, collaborative agents, and multimodal document understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991808,
"input": 991808,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.17,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/alibaba/qwen3.8-max-0902\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.8-max-0902\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.6-flash": {
"id": "alibaba/qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991808,
"input": 991808,
"output": 65536
},
"cost": {
"input": 0.19,
"output": 1.16,
"cache_read": 0.02,
"cache_write": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/alibaba/qwen3.6-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.8-flash": {
"id": "alibaba/qwen3.8-flash",
"name": "Qwen3.8 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991808,
"input": 991808,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.42,
"cache_read": 0.016,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/alibaba/qwen3.8-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-ultra-550b-a55b:thinking": {
"id": "nvidia/nemotron-3-ultra-550b-a55b:thinking",
"name": "Nvidia Nemotron 3 Ultra 550B Thinking",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 2.5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nvidia/nemotron-3-ultra-550b-a55b:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-ultra-550b-a55b:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3.5-lightning": {
"id": "nvidia/nemotron-3.5-lightning",
"name": "Nvidia Nemotron 3.5 Lightning",
"description": "Fast NVIDIA Nemotron MoE for reliable agentic tasks across enterprise workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-11",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 65536
},
"cost": {
"input": 0.05,
"output": 0.2,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nvidia/nemotron-3.5-lightning\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3.5-lightning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-super-120b-a12b:thinking": {
"id": "nvidia/nemotron-3-super-120b-a12b:thinking",
"name": "Nvidia Nemotron 3 Super 120B Thinking",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-01",
"last_updated": "2026-03-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.05,
"output": 0.25,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nvidia/nemotron-3-super-120b-a12b:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-super-120b-a12b:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3.5-lightning:thinking": {
"id": "nvidia/nemotron-3.5-lightning:thinking",
"name": "Nvidia Nemotron 3.5 Lightning Thinking",
"description": "Fast NVIDIA Nemotron MoE for reliable agentic tasks across enterprise workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-11",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 65536
},
"cost": {
"input": 0.05,
"output": 0.2,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nvidia/nemotron-3.5-lightning:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3.5-lightning:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-super-120b-a12b": {
"id": "nvidia/nemotron-3-super-120b-a12b",
"name": "Nvidia Nemotron 3 Super 120B",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-01",
"last_updated": "2026-03-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 16384
},
"cost": {
"input": 0.05,
"output": 0.25,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nvidia/nemotron-3-super-120b-a12b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-super-120b-a12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Llama-3.1-Nemotron-70B-Instruct-HF": {
"id": "nvidia/Llama-3.1-Nemotron-70B-Instruct-HF",
"name": "Nvidia Nemotron 70b",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-04-15",
"last_updated": "2025-04-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.357,
"output": 0.408,
"cache_read": 0.1785
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nvidia/Llama-3.1-Nemotron-70B-Instruct-HF\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Llama-3.1-Nemotron-70B-Instruct-HF\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-ultra-550b-a55b": {
"id": "nvidia/nemotron-3-ultra-550b-a55b",
"name": "Nvidia Nemotron 3 Ultra 550B",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 2.5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nvidia/nemotron-3-ultra-550b-a55b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-ultra-550b-a55b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Llama-3.3-Nemotron-Super-49B-v1": {
"id": "nvidia/Llama-3.3-Nemotron-Super-49B-v1",
"name": "Nvidia Nemotron Super 49B",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-08",
"last_updated": "2025-08-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nvidia/Llama-3.3-Nemotron-Super-49B-v1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Llama-3.3-Nemotron-Super-49B-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-nano-30b-a3b": {
"id": "nvidia/nemotron-3-nano-30b-a3b",
"name": "Nvidia Nemotron 3 Nano 30B",
"description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-15",
"last_updated": "2025-12-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 235929
},
"cost": {
"input": 0.17,
"output": 0.68,
"cache_read": 0.085
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nvidia/nemotron-3-nano-30b-a3b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-nano-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.8": {
"id": "anthropic/claude-opus-4.8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-opus-4.8\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.6:thinking": {
"id": "anthropic/claude-opus-4.6:thinking",
"name": "Claude 4.6 Opus Thinking",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-opus-4.6:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.6:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-latest": {
"id": "anthropic/claude-fable-latest",
"name": "Claude Fable Latest",
"description": "Compatibility alias for Claude Fable.",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-fable-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.7": {
"id": "anthropic/claude-opus-4.7",
"name": "Claude 4.7 Opus",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-opus-4.7\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-5": {
"id": "anthropic/claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-opus-5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.6": {
"id": "anthropic/claude-sonnet-4.6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-sonnet-4.6\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.6:thinking:low": {
"id": "anthropic/claude-opus-4.6:thinking:low",
"name": "Claude 4.6 Opus Thinking Low",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-opus-4.6:thinking:low\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.6:thinking:low\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-latest": {
"id": "anthropic/claude-opus-latest",
"name": "Claude Opus Latest",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-29",
"last_updated": "2026-03-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-opus-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.6:thinking:medium": {
"id": "anthropic/claude-opus-4.6:thinking:medium",
"name": "Claude 4.6 Opus Thinking Medium",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-opus-4.6:thinking:medium\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.6:thinking:medium\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-5:thinking": {
"id": "anthropic/claude-sonnet-5:thinking",
"name": "Claude Sonnet 5 Thinking",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-sonnet-5:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-5:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.6": {
"id": "anthropic/claude-opus-4.6",
"name": "Claude 4.6 Opus",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-opus-4.6\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5": {
"id": "anthropic/claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-fable-5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-latest": {
"id": "anthropic/claude-haiku-latest",
"name": "Claude Haiku Latest",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-29",
"last_updated": "2026-03-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-haiku-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.6:thinking": {
"id": "anthropic/claude-sonnet-4.6:thinking",
"name": "Claude Sonnet 4.6 Thinking",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-sonnet-4.6:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.6:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-latest": {
"id": "anthropic/claude-sonnet-latest",
"name": "Claude Sonnet Latest",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-01",
"last_updated": "2026-03-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-sonnet-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.8:thinking": {
"id": "anthropic/claude-opus-4.8:thinking",
"name": "Claude Opus 4.8 Thinking",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-opus-4.8:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.8:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.6:thinking:max": {
"id": "anthropic/claude-opus-4.6:thinking:max",
"name": "Claude 4.6 Opus Thinking Max",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-opus-4.6:thinking:max\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.6:thinking:max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.7:thinking": {
"id": "anthropic/claude-opus-4.7:thinking",
"name": "Claude 4.7 Opus Thinking",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-opus-4.7:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.7:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-5": {
"id": "anthropic/claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-sonnet-5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5.1": {
"id": "anthropic/claude-fable-5.1",
"name": "Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthropic/claude-fable-5.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-26b-a4b-it": {
"id": "google/gemma-4-26b-a4b-it",
"name": "Gemma 4 26B A4B",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 131072
},
"cost": {
"input": 0.12,
"output": 0.38,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-pro-latest": {
"id": "google/gemini-pro-latest",
"name": "Gemini Pro Latest",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-29",
"last_updated": "2026-03-29",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-pro-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-pro-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-flash-thinking": {
"id": "google/gemini-3.5-flash-thinking",
"name": "Gemini 3.5 Flash Thinking",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"cache_write": 0.083333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3.5-flash-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-flash-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3-flash-preview-thinking": {
"id": "google/gemini-3-flash-preview-thinking",
"name": "Gemini 3 Flash Thinking",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3-flash-preview-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3-flash-preview-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-pro-preview-customtools": {
"id": "google/gemini-3.1-pro-preview-customtools",
"name": "Gemini 3.1 Pro (Preview Custom Tools)",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3.1-pro-preview-customtools\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-pro-preview-customtools\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-pro-preview": {
"id": "google/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro (Preview)",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.6-flash": {
"id": "google/gemini-3.6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.041667
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3.6-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-flash-lite": {
"id": "google/gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-flash": {
"id": "google/gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"cache_write": 0.083333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3.5-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-pro-preview-high": {
"id": "google/gemini-3.1-pro-preview-high",
"name": "Gemini 3.1 Pro (Preview High)",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-02-21",
"last_updated": "2026-02-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3.1-pro-preview-high\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-pro-preview-high\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-flash-lite": {
"id": "google/gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31b-it:thinking": {
"id": "google/gemma-4-31b-it:thinking",
"name": "Gemma 4 31B Thinking",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.35,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemma-4-31b-it:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31b-it:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-flash-lite-latest": {
"id": "google/gemini-flash-lite-latest",
"name": "Gemini Flash Lite Latest",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-flash-lite-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-flash-lite-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31b-it": {
"id": "google/gemma-4-31b-it",
"name": "Gemma 4 31B",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.45,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemma-4-31b-it\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-26b-a4b-it:thinking": {
"id": "google/gemma-4-26b-a4b-it:thinking",
"name": "Gemma 4 26B A4B Thinking",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 131072
},
"cost": {
"input": 0.13,
"output": 0.4,
"cache_read": 0.065
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemma-4-26b-a4b-it:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-26b-a4b-it:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3-flash-preview": {
"id": "google/gemini-3-flash-preview",
"name": "Gemini 3 Flash (Preview)",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3-flash-preview\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.8-flash": {
"id": "google/gemini-3.8-flash",
"name": "Gemini 3.8 Flash",
"description": "Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.041667
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3.8-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.7-flash": {
"id": "google/gemini-3.7-flash",
"name": "Gemini 3.7 Flash",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.041667
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3.7-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-flash-latest": {
"id": "google/gemini-flash-latest",
"name": "Gemini Flash Latest",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-flash-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-flash-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-pro-preview-low": {
"id": "google/gemini-3.1-pro-preview-low",
"name": "Gemini 3.1 Pro (Preview Low)",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-02-21",
"last_updated": "2026-02-21",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/google/gemini-3.1-pro-preview-low\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-pro-preview-low\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baseten/Kimi-K2-Instruct-FP4": {
"id": "baseten/Kimi-K2-Instruct-FP4",
"name": "Kimi K2 0711 Instruct FP4",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-07-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 131072
},
"cost": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/baseten/Kimi-K2-Instruct-FP4\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"baseten/Kimi-K2-Instruct-FP4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"crofai/greg-2-super": {
"id": "crofai/greg-2-super",
"name": "Greg 2 Super",
"description": "Greg 2 Super is CrofAI's balanced Greg 2 model for strong UI design, frontend iteration, coding, writing, and everyday agent tasks at a lower cost than Ultra.",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-06-19",
"last_updated": "2026-06-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 229376,
"input": 229376,
"output": 229376
},
"cost": {
"input": 1.5,
"output": 5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/crofai/greg-2-super\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"crofai/greg-2-super\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"crofai/greg-2-ultra": {
"id": "crofai/greg-2-ultra",
"name": "Greg 2 Ultra",
"description": "Greg 2 Ultra is CrofAI's most capable Greg 2 model, tuned for premium UI design, agentic coding, creative writing, and higher-end general reasoning tasks.",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-06-19",
"last_updated": "2026-06-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 229376,
"input": 229376,
"output": 229376
},
"cost": {
"input": 3,
"output": 10,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/crofai/greg-2-ultra\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"crofai/greg-2-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/Inkling-Small": {
"id": "thinkingmachines/Inkling-Small",
"name": "Inkling Small",
"description": "Multimodal MoE reasoning model (276B total, 12B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-30",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"input": 524288,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 1.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/thinkingmachines/Inkling-Small\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/Inkling-Small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/inkling:thinking": {
"id": "thinkingmachines/inkling:thinking",
"name": "Inkling Thinking",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048000,
"input": 1048000,
"output": 32768
},
"cost": {
"input": 1,
"output": 4.05,
"cache_read": 0.17
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/thinkingmachines/inkling:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/inkling:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/inkling": {
"id": "thinkingmachines/inkling",
"name": "Inkling",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048000,
"input": 1048000,
"output": 32768
},
"cost": {
"input": 1,
"output": 4.05,
"cache_read": 0.17
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/thinkingmachines/inkling\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/Inkling-Small:thinking": {
"id": "thinkingmachines/Inkling-Small:thinking",
"name": "Inkling Small Thinking",
"description": "Multimodal MoE reasoning model (276B total, 12B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-30",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"input": 524288,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 1.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/thinkingmachines/Inkling-Small:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/Inkling-Small:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.3": {
"id": "meta/muse-spark-1.3",
"name": "Muse Spark 1.3",
"description": "Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It improves long-horizon agent collaboration, instruction following, and coding efficiency relative to Muse Spark 1.2.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 943718
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta/muse-spark-1.3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.2": {
"id": "meta/muse-spark-1.2",
"name": "Muse Spark 1.2",
"description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta/muse-spark-1.2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.2-contributor": {
"id": "meta/muse-spark-1.2-contributor",
"name": "Muse Spark 1.2 Contributor (Data Used for Training)",
"description": "A much cheaper opt-in version of Muse Spark 1.2 with the same multimodal coding and agentic capabilities. Prompts and outputs sent to this Contributor model may be used by Meta for training and to improve its products; use the standard Muse Spark 1.2 model if you do not want your data used for training.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.002
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta/muse-spark-1.2-contributor\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.2-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.3-contributor": {
"id": "meta/muse-spark-1.3-contributor",
"name": "Muse Spark 1.3 Contributor",
"description": "Meta's Muse Spark 1.3 Contributor is a frontier multimodal reasoning model for long-horizon coding and agentic workflows, with strong gains in computer use, browsing, professional tool use, codebase understanding, and million-token retrieval. It accepts text, images, audio, video, and files, supports tool calling and structured output, and always reasons before answering. Prompts and outputs may be used by Meta for training and to improve its products.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 943718
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.002
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta/muse-spark-1.3-contributor\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.3-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.1": {
"id": "meta/muse-spark-1.1",
"name": "Muse Spark 1.1",
"description": "Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-08",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta/muse-spark-1.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-glimmer-30b": {
"id": "meta/muse-glimmer-30b",
"name": "Muse Glimmer 30B",
"description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-01-04",
"release_date": "2026-08-10",
"last_updated": "2026-08-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 117964
},
"cost": {
"input": 0.35,
"output": 1.5,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta/muse-glimmer-30b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-glimmer-30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perceptron/perceptron-mk1": {
"id": "perceptron/perceptron-mk1",
"name": "Perceptron Mk1",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": true,
"release_date": "2026-05-12",
"last_updated": "2026-05-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"input": 32768,
"output": 8192
},
"cost": {
"input": 0.15,
"output": 1.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/perceptron/perceptron-mk1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"perceptron/perceptron-mk1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Steelskull/L3.3-Cu-Mai-R1-70b": {
"id": "Steelskull/L3.3-Cu-Mai-R1-70b",
"name": "Llama 3.3 70B Cu Mai",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Steelskull/L3.3-Cu-Mai-R1-70b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Steelskull/L3.3-Cu-Mai-R1-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Steelskull/L3.3-Electra-R1-70b": {
"id": "Steelskull/L3.3-Electra-R1-70b",
"name": "Steelskull Electra R1 70b",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 0.69989,
"output": 0.69989,
"cache_read": 0.349945
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Steelskull/L3.3-Electra-R1-70b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Steelskull/L3.3-Electra-R1-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Steelskull/L3.3-Nevoria-R1-70b": {
"id": "Steelskull/L3.3-Nevoria-R1-70b",
"name": "Steelskull Nevoria R1 70b",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Steelskull/L3.3-Nevoria-R1-70b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Steelskull/L3.3-Nevoria-R1-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Steelskull/L3.3-MS-Nevoria-70b": {
"id": "Steelskull/L3.3-MS-Nevoria-70b",
"name": "Steelskull Nevoria 70b",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Steelskull/L3.3-MS-Nevoria-70b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Steelskull/L3.3-MS-Nevoria-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/doubao-seed-2.1-turbo": {
"id": "bytedance/doubao-seed-2.1-turbo",
"name": "Doubao Seed 2.1 Turbo",
"description": "Fast, lower-cost model in the Doubao Seed 2.1 family for everyday chat, coding assistance, document work, and high-throughput productivity tasks. Supports a 256k context window and up to 128k output tokens. Note: privacy and logging guarantees may be limited.",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-06-23",
"last_updated": "2026-06-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 128000
},
"cost": {
"input": 0.5,
"output": 2.5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/bytedance/doubao-seed-2.1-turbo\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/doubao-seed-2.1-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/doubao-seed-2.1-pro": {
"id": "bytedance/doubao-seed-2.1-pro",
"name": "Doubao Seed 2.1 Pro",
"description": "Higher-capability model in the Doubao Seed 2.1 family for agentic coding, long-context analysis, complex instruction following, and productivity workflows. Supports a 256k context window and up to 128k output tokens. Note: privacy and logging guarantees may be limited.",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-06-23",
"last_updated": "2026-06-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 128000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/bytedance/doubao-seed-2.1-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/doubao-seed-2.1-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/doubao-seed-character": {
"id": "bytedance/doubao-seed-character",
"name": "Doubao Seed Character",
"description": "ByteDance's character-focused Doubao Seed model for roleplay, persona consistency, dialogue, and creative character interactions. It supports text and image input with a 128k context window. Requests route through ZenMux to ByteDance; ZenMux does not publish a model-API zero-retention or training guarantee, so avoid sensitive data.",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"release_date": "2026-07-18",
"last_updated": "2026-07-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 32768
},
"cost": {
"input": 0.1179,
"output": 0.2947,
"cache_read": 0.0236,
"cache_write": 0.0025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/bytedance/doubao-seed-character\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/doubao-seed-character\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance-seed/seed-2-1-turbo": {
"id": "bytedance-seed/seed-2-1-turbo",
"name": "ByteDance Seed 2.1 Turbo",
"description": "ByteDance Seed 2.1 Turbo is a multimodal model for coding and long-horizon agent workflows, including end-to-end software delivery and multi-step task execution. It supports text, image, and video input with a 262k context window.",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"input": 262144,
"output": 235929
},
"cost": {
"input": 0.5,
"output": 2.5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/bytedance-seed/seed-2-1-turbo\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"bytedance-seed/seed-2-1-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance-seed/seed-2.0-code": {
"id": "bytedance-seed/seed-2.0-code",
"name": "ByteDance Seed 2.0 Code",
"description": "ByteDance Seed coding model for multimodal software engineering and long-running agents",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"input": 262144,
"output": 131072
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/bytedance-seed/seed-2.0-code\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"bytedance-seed/seed-2.0-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance-seed/seed-2.0-lite": {
"id": "bytedance-seed/seed-2.0-lite",
"name": "ByteDance Seed 2.0 Lite",
"description": "Cost-efficient ByteDance Seed 2.0 model for production chat, analysis, and structured generation",
"family": "seed",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"input": 262144,
"output": 131072
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/bytedance-seed/seed-2.0-lite\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"bytedance-seed/seed-2.0-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"NeverSleep/Lumimaid-v0.2-70B": {
"id": "NeverSleep/Lumimaid-v0.2-70B",
"name": "Lumimaid v0.2",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 1,
"output": 1.5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/NeverSleep/Lumimaid-v0.2-70B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"NeverSleep/Lumimaid-v0.2-70B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/qwen3.5-27b": {
"id": "TEE/qwen3.5-27b",
"name": "Qwen3.5 27B TEE",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.4,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/qwen3.5-27b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/qwen3.5-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/qwen3.8-27b": {
"id": "TEE/qwen3.8-27b",
"name": "Qwen3.8 27B TEE",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 262144
},
"cost": {
"input": 0.4,
"output": 3,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/qwen3.8-27b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/kimi-k2.6": {
"id": "TEE/kimi-k2.6",
"name": "Kimi K2.6 TEE",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 5.25,
"cache_read": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/kimi-k2.6\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/nemotron-3.5-lightning": {
"id": "TEE/nemotron-3.5-lightning",
"name": "Nvidia Nemotron 3.5 Lightning TEE",
"description": "Fast NVIDIA Nemotron MoE for reliable agentic tasks across enterprise workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-11",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 0.08,
"output": 0.2,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/nemotron-3.5-lightning\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/nemotron-3.5-lightning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/glm-5.2": {
"id": "TEE/glm-5.2",
"name": "GLM 5.2 TEE",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/glm-5.2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/gemma4-31b": {
"id": "TEE/gemma4-31b",
"name": "Gemma 4 31B",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-04-04",
"last_updated": "2026-04-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 131072
},
"cost": {
"input": 0.4,
"output": 1,
"cache_read": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/gemma4-31b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/gemma4-31b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/deepseek-v4-flash": {
"id": "TEE/deepseek-v4-flash",
"name": "DeepSeek V4 Flash TEE",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 393216
},
"cost": {
"input": 0.2,
"output": 0.4,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/deepseek-v4-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/kimi-k2.7-code": {
"id": "TEE/kimi-k2.7-code",
"name": "Kimi K2.7 Code TEE",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/kimi-k2.7-code\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/gpt-oss-20b": {
"id": "TEE/gpt-oss-20b",
"name": "GPT-OSS 20B TEE",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 8192
},
"cost": {
"input": 0.04,
"output": 0.15,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/gpt-oss-20b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/qwen2.5-vl-72b-instruct": {
"id": "TEE/qwen2.5-vl-72b-instruct",
"name": "Qwen2.5 VL 72B TEE",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-02-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"input": 65536,
"output": 8192
},
"cost": {
"input": 0.7,
"output": 0.7,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/qwen2.5-vl-72b-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/qwen2.5-vl-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/gemma4-31b:thinking": {
"id": "TEE/gemma4-31b:thinking",
"name": "Gemma 4 31B Thinking TEE",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-02",
"last_updated": "2026-05-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 131072
},
"cost": {
"input": 0.4,
"output": 1,
"cache_read": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/gemma4-31b:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/gemma4-31b:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/qwen3.5-397b-a17b": {
"id": "TEE/qwen3.5-397b-a17b",
"name": "Qwen3.5 397B A17B TEE",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 0.55,
"output": 3.5,
"cache_read": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/gemma-4-26b-a4b-uncensored": {
"id": "TEE/gemma-4-26b-a4b-uncensored",
"name": "Gemma 4 26B A4B Uncensored TEE",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-23",
"last_updated": "2026-05-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"input": 65536,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 0.7,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/gemma-4-26b-a4b-uncensored\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/gemma-4-26b-a4b-uncensored\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/qwen3.6-27b": {
"id": "TEE/qwen3.6-27b",
"name": "Qwen3.6 27B TEE",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 0.32,
"output": 2.7,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/qwen3.6-27b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/qwen3.6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/kimi-k3": {
"id": "TEE/kimi-k3",
"name": "Kimi K3 TEE",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65535
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/kimi-k3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/deepseek-v3.2": {
"id": "TEE/deepseek-v3.2",
"name": "DeepSeek V3.2 TEE",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 164000,
"input": 164000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 1,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/deepseek-v3.2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/qwen3.6-35b-a3b": {
"id": "TEE/qwen3.6-35b-a3b",
"name": "Qwen3.6 35B A3B TEE",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 262144
},
"cost": {
"input": 0.2,
"output": 1.27,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/qwen3.6-35b-a3b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/glm-5.3-flash": {
"id": "TEE/glm-5.3-flash",
"name": "GLM 5.3 Flash TEE",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/glm-5.3-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/muse-glimmer-30b": {
"id": "TEE/muse-glimmer-30b",
"name": "Muse Glimmer 30B TEE",
"description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-01-04",
"release_date": "2026-08-10",
"last_updated": "2026-08-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 131072
},
"cost": {
"input": 0.35,
"output": 1.5,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/muse-glimmer-30b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/muse-glimmer-30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/gemma-4-31b-it": {
"id": "TEE/gemma-4-31b-it",
"name": "Gemma 4 31B IT TEE",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 262144
},
"cost": {
"input": 0.15,
"output": 0.46,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/gemma-4-31b-it\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/qwen3.6-35b-a3b-uncensored": {
"id": "TEE/qwen3.6-35b-a3b-uncensored",
"name": "Qwen3.6 35B A3B Uncensored TEE",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024,
"max": 131072
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-23",
"last_updated": "2026-05-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/qwen3.6-35b-a3b-uncensored\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/qwen3.6-35b-a3b-uncensored\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/qwen3-8b": {
"id": "TEE/qwen3-8b",
"name": "Qwen3 8B TEE",
"description": "Qwen3 8B is an open-weight dense language model from Alibaba's Qwen team for efficient dialogue, reasoning, mathematics, coding, and tool use. Running inside a TEE (Trusted Execution Environment), with provider attestation support.",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"input": 40960,
"output": 8192
},
"cost": {
"input": 0.11,
"output": 0.45,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/qwen3-8b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/qwen3-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/glm-5.1": {
"id": "TEE/glm-5.1",
"name": "GLM 5.1 TEE",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"input": 202752,
"output": 65535
},
"cost": {
"input": 1.5,
"output": 5.25,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/glm-5.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/glm-5.2:thinking": {
"id": "TEE/glm-5.2:thinking",
"name": "GLM 5.2 Thinking TEE",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/glm-5.2:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/glm-5.2:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/gpt-oss-120b": {
"id": "TEE/gpt-oss-120b",
"name": "GPT-OSS 120B TEE",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 16384
},
"cost": {
"input": 2,
"output": 2,
"cache_read": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/gpt-oss-120b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/llama3-3-70b": {
"id": "TEE/llama3-3-70b",
"name": "Llama 3.3 70B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-03",
"last_updated": "2025-07-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 1.75,
"output": 2.75,
"cache_read": 1.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/llama3-3-70b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/llama3-3-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/glm-5.1-thinking": {
"id": "TEE/glm-5.1-thinking",
"name": "GLM 5.1 Thinking TEE",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"input": 202752,
"output": 65535
},
"cost": {
"input": 1.5,
"output": 5.25,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/glm-5.1-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/glm-5.1-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TEE/glm-5.3": {
"id": "TEE/glm-5.3",
"name": "GLM 5.3 TEE",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TEE/glm-5.3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TEE/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meganova-ai/manta-mini-1.0": {
"id": "meganova-ai/manta-mini-1.0",
"name": "Manta Mini 1.0",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "nova",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-20",
"last_updated": "2025-12-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"input": 8192,
"output": 8192
},
"cost": {
"input": 0.02,
"output": 0.16,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meganova-ai/manta-mini-1.0\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meganova-ai/manta-mini-1.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meganova-ai/manta-flash-1.0": {
"id": "meganova-ai/manta-flash-1.0",
"name": "Manta Flash 1.0",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "nova",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-20",
"last_updated": "2025-12-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"input": 16384,
"output": 16384
},
"cost": {
"input": 0.02,
"output": 0.16,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meganova-ai/manta-flash-1.0\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meganova-ai/manta-flash-1.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meganova-ai/manta-pro-1.0": {
"id": "meganova-ai/manta-pro-1.0",
"name": "Manta Pro 1.0",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "nova",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-20",
"last_updated": "2025-12-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"input": 65536,
"output": 32768
},
"cost": {
"input": 0.06,
"output": 0.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meganova-ai/manta-pro-1.0\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meganova-ai/manta-pro-1.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inception/mercury-2.5-preview": {
"id": "inception/mercury-2.5-preview",
"name": "Mercury 2.5 Preview",
"description": "Mercury 2.5 Preview is Inception's latest and most intelligent diffusion language model. Instead of generating tokens strictly one at a time, it produces and refines multiple tokens in parallel, reaching up to 1,107 tokens per second on standard GPUs. It delivers a 10+ point intelligence gain over Mercury 2, with tunable reasoning, parallel tool calls, schema-aligned JSON output, and a 260K context window. It is built for latency-sensitive production work such as search agents, voice pipelines, customer support, rapid coding iteration, and coding subagents.",
"family": "mercury",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 260000,
"input": 260000,
"output": 65536
},
"cost": {
"input": 0.04,
"output": 0.15,
"cache_read": 0.004
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/inception/mercury-2.5-preview\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"inception/mercury-2.5-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"unsloth/gemma-3-4b-it": {
"id": "unsloth/gemma-3-4b-it",
"name": "Gemma 3 4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "unsloth",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-03-10",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 8192
},
"cost": {
"input": 0.2006,
"output": 0.2006,
"cache_read": 0.1003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/unsloth/gemma-3-4b-it\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"unsloth/gemma-3-4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"unsloth/gemma-3-27b-it": {
"id": "unsloth/gemma-3-27b-it",
"name": "Gemma 3 27B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "unsloth",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-03-10",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 96000
},
"cost": {
"input": 0.2992,
"output": 0.2992,
"cache_read": 0.1496
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/unsloth/gemma-3-27b-it\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"unsloth/gemma-3-27b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"unsloth/gemma-3-12b-it": {
"id": "unsloth/gemma-3-12b-it",
"name": "Gemma 3 12B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "unsloth",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-03-10",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 16384
},
"cost": {
"input": 0.272,
"output": 0.272,
"cache_read": 0.136
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/unsloth/gemma-3-12b-it\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"unsloth/gemma-3-12b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"microsoft/wizardlm-2-8x22b": {
"id": "microsoft/wizardlm-2-8x22b",
"name": "WizardLM-2 8x22B",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-04-15",
"last_updated": "2025-04-15",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"input": 65536,
"output": 8192
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/microsoft/wizardlm-2-8x22b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"microsoft/wizardlm-2-8x22b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inflatebot/MN-12B-Mag-Mell-R1": {
"id": "inflatebot/MN-12B-Mag-Mell-R1",
"name": "Mag Mell R1",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-nemo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/inflatebot/MN-12B-Mag-Mell-R1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"inflatebot/MN-12B-Mag-Mell-R1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepcogito/cogito-v1-preview-qwen-32B": {
"id": "deepcogito/cogito-v1-preview-qwen-32B",
"name": "Cogito v1 Preview Qwen 32B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-05-10",
"last_updated": "2025-05-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 32768
},
"cost": {
"input": 1.8,
"output": 1.8,
"cache_read": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepcogito/cogito-v1-preview-qwen-32B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepcogito/cogito-v1-preview-qwen-32B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sakana/fugu-ultra": {
"id": "sakana/fugu-ultra",
"name": "Fugu Ultra",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 16384
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/sakana/fugu-ultra\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"sakana/fugu-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sakana/fugu-max": {
"id": "sakana/fugu-max",
"name": "Fugu Max",
"description": "Sakana AI's cost-performance Fugu model uses learned multi-agent orchestration to route tasks across expert models for reasoning, coding, and tool use.",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-11",
"last_updated": "2026-09-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/sakana/fugu-max\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"sakana/fugu-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sakana/fugu-ultra-v1.1": {
"id": "sakana/fugu-ultra-v1.1",
"name": "Fugu Ultra v1.1",
"description": "Sakana AI's upgraded Fugu Ultra release with stronger coding, agentic task execution, and advanced reasoning through dynamic orchestration of frontier models.",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 16384
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/sakana/fugu-ultra-v1.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"sakana/fugu-ultra-v1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Envoid/Llama-3.05-Nemotron-Tenyxchat-Storybreaker-70B": {
"id": "Envoid/Llama-3.05-Nemotron-Tenyxchat-Storybreaker-70B",
"name": "Nemotron Tenyxchat Storybreaker 70b",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Envoid/Llama-3.05-Nemotron-Tenyxchat-Storybreaker-70B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Envoid/Llama-3.05-Nemotron-Tenyxchat-Storybreaker-70B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Envoid/Llama-3.05-NT-Storybreaker-Ministral-70B": {
"id": "Envoid/Llama-3.05-NT-Storybreaker-Ministral-70B",
"name": "Llama 3.05 Storybreaker Ministral 70b",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Envoid/Llama-3.05-NT-Storybreaker-Ministral-70B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Envoid/Llama-3.05-NT-Storybreaker-Ministral-70B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"NousResearch/hermes-3-llama-3.1-70b": {
"id": "NousResearch/hermes-3-llama-3.1-70b",
"name": "Hermes 3 70B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "nousresearch",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2026-01-07",
"last_updated": "2026-01-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"input": 65536,
"output": 8192
},
"cost": {
"input": 0.408,
"output": 0.408,
"cache_read": 0.204
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/NousResearch/hermes-3-llama-3.1-70b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"NousResearch/hermes-3-llama-3.1-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"NousResearch/hermes-4-405b": {
"id": "NousResearch/hermes-4-405b",
"name": "Hermes 4 Large",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "nousresearch",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"release_date": "2025-08-26",
"last_updated": "2025-08-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 8192
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/NousResearch/hermes-4-405b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"NousResearch/hermes-4-405b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"NousResearch/hermes-4-405b:thinking": {
"id": "NousResearch/hermes-4-405b:thinking",
"name": "Hermes 4 Large (Thinking)",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "nousresearch",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 8192
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/NousResearch/hermes-4-405b:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"NousResearch/hermes-4-405b:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"failspy/Meta-Llama-3-70B-Instruct-abliterated-v3.5": {
"id": "failspy/Meta-Llama-3-70B-Instruct-abliterated-v3.5",
"name": "Llama 3 70B abliterated",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-26",
"last_updated": "2025-07-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"input": 8192,
"output": 8192
},
"cost": {
"input": 0.7,
"output": 0.7,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/failspy/Meta-Llama-3-70B-Instruct-abliterated-v3.5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"failspy/Meta-Llama-3-70B-Instruct-abliterated-v3.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"GalrionSoftworks/MN-LooseCannon-12B-v1": {
"id": "GalrionSoftworks/MN-LooseCannon-12B-v1",
"name": "MN-LooseCannon-12B-v1",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-nemo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/GalrionSoftworks/MN-LooseCannon-12B-v1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"GalrionSoftworks/MN-LooseCannon-12B-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ibm-granite/granite-4.2-8b": {
"id": "ibm-granite/granite-4.2-8b",
"name": "Granite 4.2 8B",
"description": "IBM Granite 4.2 8B is an Apache 2.0-licensed dense model with native step-by-step reasoning and specialized training for agentic work. It can plan before acting, sequence tools, navigate codebases, work in terminals, and verify results across coding, search, mathematics, science, and complex instruction-following tasks.",
"family": "granite",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-31",
"last_updated": "2026-08-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 117964
},
"cost": {
"input": 0.1,
"output": 0.15,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/ibm-granite/granite-4.2-8b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"ibm-granite/granite-4.2-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-vision-exp": {
"id": "deepseek/deepseek-v4-flash-vision-exp",
"name": "DeepSeek V4 Flash Vision Exp",
"description": "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-21",
"last_updated": "2026-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 384000
},
"cost": {
"input": 0.22,
"output": 0.66,
"cache_read": 0.007
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4-flash-vision-exp\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-vision-exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro-0813": {
"id": "deepseek/deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 384000
},
"cost": {
"input": 1.1,
"output": 2.5,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-0731": {
"id": "deepseek/deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.16,
"cache_read": 0.013
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-latest": {
"id": "deepseek/deepseek-v4-flash-latest",
"name": "DeepSeek V4 Flash Latest",
"description": "Compatibility alias that routes to the newest dated DeepSeek V4 Flash release. Currently routes to DeepSeek V4 Flash 0731. ⚠️ This route goes directly to DeepSeek, so privacy and logging guarantees are limited.",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-02",
"last_updated": "2026-08-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 384000
},
"cost": {
"input": 0.05,
"output": 0.16,
"cache_read": 0.013
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4-flash-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash": {
"id": "deepseek/deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 384000
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-0731:thinking": {
"id": "deepseek/deepseek-v4-flash-0731:thinking",
"name": "DeepSeek V4 Flash 0731 (Thinking)",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.16,
"cache_read": 0.013
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4-flash-0731:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-0731:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash:thinking": {
"id": "deepseek/deepseek-v4-flash:thinking",
"name": "DeepSeek V4 Flash (Thinking)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 384000
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4-flash:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4.1-flash": {
"id": "deepseek/deepseek-v4.1-flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 384000
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4.1-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4.1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro-0813:thinking": {
"id": "deepseek/deepseek-v4-pro-0813:thinking",
"name": "DeepSeek V4 Pro 0813 Thinking",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 384000
},
"cost": {
"input": 1.1,
"output": 2.5,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4-pro-0813:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro-0813:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4.1-flash:thinking": {
"id": "deepseek/deepseek-v4.1-flash:thinking",
"name": "DeepSeek V4.1 Flash Thinking",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 384000
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4.1-flash:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4.1-flash:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro:thinking": {
"id": "deepseek/deepseek-v4-pro:thinking",
"name": "DeepSeek V4 Pro (Thinking)",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 384000
},
"cost": {
"input": 1.1,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4-pro:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.2": {
"id": "deepseek/deepseek-v3.2",
"name": "DeepSeek V3.2",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use",
"family": "deepseek",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163000,
"input": 163000,
"output": 65536
},
"cost": {
"input": 0.28,
"output": 0.42,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v3.2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-latest": {
"id": "deepseek/deepseek-latest",
"name": "DeepSeek Latest",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-03",
"last_updated": "2026-05-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 384000
},
"cost": {
"input": 1.1,
"output": 2.5,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro": {
"id": "deepseek/deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 384000
},
"cost": {
"input": 1.1,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v4-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.2:thinking": {
"id": "deepseek/deepseek-v3.2:thinking",
"name": "DeepSeek V3.2 Thinking",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use",
"family": "deepseek",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163000,
"input": 163000,
"output": 65536
},
"cost": {
"input": 0.28,
"output": 0.42,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/deepseek/deepseek-v3.2:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.2:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"EVA-UNIT-01/EVA-LLaMA-3.33-70B-v0.0": {
"id": "EVA-UNIT-01/EVA-LLaMA-3.33-70B-v0.0",
"name": "EVA Llama 3.33 70B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-26",
"last_updated": "2025-07-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 2.006,
"output": 2.006,
"cache_read": 1.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/EVA-UNIT-01/EVA-LLaMA-3.33-70B-v0.0\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"EVA-UNIT-01/EVA-LLaMA-3.33-70B-v0.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"EVA-UNIT-01/EVA-LLaMA-3.33-70B-v0.1": {
"id": "EVA-UNIT-01/EVA-LLaMA-3.33-70B-v0.1",
"name": "EVA-LLaMA-3.33-70B-v0.1",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 16384
},
"cost": {
"input": 2.006,
"output": 2.006,
"cache_read": 1.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/EVA-UNIT-01/EVA-LLaMA-3.33-70B-v0.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"EVA-UNIT-01/EVA-LLaMA-3.33-70B-v0.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"EVA-UNIT-01/EVA-Qwen2.5-32B-v0.2": {
"id": "EVA-UNIT-01/EVA-Qwen2.5-32B-v0.2",
"name": "EVA-Qwen2.5-32B-v0.2",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-26",
"last_updated": "2025-07-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.799,
"output": 0.799,
"cache_read": 0.3995
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/EVA-UNIT-01/EVA-Qwen2.5-32B-v0.2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"EVA-UNIT-01/EVA-Qwen2.5-32B-v0.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"EVA-UNIT-01/EVA-Qwen2.5-72B-v0.2": {
"id": "EVA-UNIT-01/EVA-Qwen2.5-72B-v0.2",
"name": "EVA-Qwen2.5-72B-v0.2",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.799,
"output": 0.799,
"cache_read": 0.3995
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/EVA-UNIT-01/EVA-Qwen2.5-72B-v0.2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"EVA-UNIT-01/EVA-Qwen2.5-72B-v0.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon/nova-2-lite-v1": {
"id": "amazon/nova-2-lite-v1",
"name": "Amazon Nova 2 Lite",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "nova",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-12-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 65535
},
"cost": {
"input": 0.51,
"output": 4.25,
"cache_read": 0.255
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/amazon/nova-2-lite-v1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"amazon/nova-2-lite-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon/nova-pro-v1": {
"id": "amazon/nova-pro-v1",
"name": "Amazon Nova Pro 1.0",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "nova-pro",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-12-03",
"last_updated": "2024-12-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 300000,
"input": 300000,
"output": 32000
},
"cost": {
"input": 0.799,
"output": 3.196,
"cache_read": 0.3995
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/amazon/nova-pro-v1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"amazon/nova-pro-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon/nova-lite-v1": {
"id": "amazon/nova-lite-v1",
"name": "Amazon Nova Lite 1.0",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "nova-lite",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-12-03",
"last_updated": "2024-12-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 300000,
"input": 300000,
"output": 5120
},
"cost": {
"input": 0.0595,
"output": 0.238,
"cache_read": 0.02975
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/amazon/nova-lite-v1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"amazon/nova-lite-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"LLM360/K2-Think": {
"id": "LLM360/K2-Think",
"name": "K2-Think",
"description": "Kimi reasoning model for long-horizon research, planning, and tool use",
"family": "kimi-thinking",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-26",
"last_updated": "2025-07-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 32768
},
"cost": {
"input": 0.17,
"output": 0.68,
"cache_read": 0.085
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/LLM360/K2-Think\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"LLM360/K2-Think\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionai/ling-3.0-flash": {
"id": "inclusionai/ling-3.0-flash",
"name": "Ling 3.0 Flash",
"description": "Ling-3.0-flash is a 124B-parameter Mixture-of-Experts model with approximately 5.1B parameters active per token. It prioritizes token efficiency and production-scale agentic inference, helping coding and tool-using agents complete more work within constrained latency and serving budgets.",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-23",
"last_updated": "2026-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.075,
"output": 0.22,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/inclusionai/ling-3.0-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"inclusionai/ling-3.0-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionai/ling-3.0-flash:thinking": {
"id": "inclusionai/ling-3.0-flash:thinking",
"name": "Ling 3.0 Flash Thinking",
"description": "Ling-3.0-flash Thinking enables visible reasoning on inclusionAI's token-efficient 124B-parameter Mixture-of-Experts model for harder coding, tool use, planning, and production-scale agent workflows.",
"family": "ling",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-23",
"last_updated": "2026-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.075,
"output": 0.22,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/inclusionai/ling-3.0-flash:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"inclusionai/ling-3.0-flash:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionai/ling-3.0-flash-vl": {
"id": "inclusionai/ling-3.0-flash-vl",
"name": "Ling 3.0 Flash VL",
"description": "Ling 3.0 Flash VL is inclusionAI's native multimodal Mixture-of-Experts model with 124B total parameters and 5.5B active parameters per token. It combines image and video understanding with reasoning and tool use for document analysis, charts, visual verification, and interface-based agent tasks. Thinking is enabled by default and can be turned off in settings.",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-09-09",
"last_updated": "2026-09-09",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.06,
"output": 0.18,
"cache_read": 0.012
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/inclusionai/ling-3.0-flash-vl\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"inclusionai/ling-3.0-flash-vl\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M1-80k": {
"id": "MiniMaxAI/MiniMax-M1-80k",
"name": "MiniMax M1 80K",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-01-08",
"last_updated": "2025-06-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 131072
},
"cost": {
"input": 0.6052,
"output": 2.4225,
"cache_read": 0.3026
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/MiniMaxAI/MiniMax-M1-80k\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M1-80k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"lightonai/LightOnOCR-2-1B": {
"id": "lightonai/LightOnOCR-2-1B",
"name": "LightOnOCR 2",
"description": "LightOnOCR 2 hosted by IONOS in Berlin, Germany. Zero data retention.",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 8192
},
"cost": {
"input": 0.1785,
"output": 0.3465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/lightonai/LightOnOCR-2-1B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"lightonai/LightOnOCR-2-1B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthracite-org/magnum-v4-72b": {
"id": "anthracite-org/magnum-v4-72b",
"name": "Magnum v4 72B",
"description": "Open Llama multimodal model for image understanding and text reasoning",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 2.006,
"output": 2.992,
"cache_read": 1.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthracite-org/magnum-v4-72b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthracite-org/magnum-v4-72b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthracite-org/magnum-v2-72b": {
"id": "anthracite-org/magnum-v2-72b",
"name": "Magnum V2 72B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 2.006,
"output": 2.992,
"cache_read": 1.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/anthracite-org/magnum-v2-72b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"anthracite-org/magnum-v2-72b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Salesforce/Llama-xLAM-2-70b-fc-r": {
"id": "Salesforce/Llama-xLAM-2-70b-fc-r",
"name": "Llama-xLAM-2 70B fc-r",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-04-13",
"last_updated": "2025-04-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 2.5,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Salesforce/Llama-xLAM-2-70b-fc-r\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Salesforce/Llama-xLAM-2-70b-fc-r\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.20-multi-agent": {
"id": "x-ai/grok-4.20-multi-agent",
"name": "Grok 4.20 Multi-Agent",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-31",
"last_updated": "2026-03-31",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"input": 2000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/x-ai/grok-4.20-multi-agent\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.20-multi-agent\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.3": {
"id": "x-ai/grok-4.3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 900000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/x-ai/grok-4.3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.20": {
"id": "x-ai/grok-4.20",
"name": "Grok 4.20",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-31",
"last_updated": "2026-03-31",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"input": 2000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/x-ai/grok-4.20\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.20\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-latest": {
"id": "x-ai/grok-latest",
"name": "Grok Latest",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-03",
"last_updated": "2026-05-03",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"input": 500000,
"output": 450000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/x-ai/grok-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.5": {
"id": "x-ai/grok-4.5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"input": 500000,
"output": 450000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/x-ai/grok-4.5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-build-0.1": {
"id": "x-ai/grok-build-0.1",
"name": "Grok Build 0.1",
"description": "Fast Grok coding model tuned for agentic engineering and iterative edits",
"family": "grok-build",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 256000,
"output": 230400
},
"cost": {
"input": 1,
"output": 2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/x-ai/grok-build-0.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-build-0.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.6": {
"id": "x-ai/grok-4.6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"input": 500000,
"output": 450000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/x-ai/grok-4.6\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-3.1-8b-instruct": {
"id": "meta-llama/llama-3.1-8b-instruct",
"name": "Llama 3.1 8b Instruct",
"description": "Compact open Llama model for lightweight chat, drafting, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 16384
},
"cost": {
"input": 0.0544,
"output": 0.085,
"cache_read": 0.0272
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta-llama/llama-3.1-8b-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-3.1-8b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-3.1-405b-instruct": {
"id": "meta-llama/llama-3.1-405b-instruct",
"name": "Llama 3.1 405B",
"description": "Llama 3.1 405B hosted by IONOS in Berlin, Germany. Zero data retention.",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 8192
},
"cost": {
"input": 2.0265,
"output": 2.0265
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta-llama/llama-3.1-405b-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-3.1-405b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-3.2-3b-instruct": {
"id": "meta-llama/llama-3.2-3b-instruct",
"name": "Llama 3.2 3b Instruct",
"description": "Open Llama multimodal model for image understanding and text reasoning",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-09-25",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 8192
},
"cost": {
"input": 0.0306,
"output": 0.0493,
"cache_read": 0.0153
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta-llama/llama-3.2-3b-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-3.2-3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-4-maverick": {
"id": "meta-llama/llama-4-maverick",
"name": "Llama 4 Maverick",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2025-09-05",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta-llama/llama-4-maverick\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-4-maverick\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-4-scout": {
"id": "meta-llama/llama-4-scout",
"name": "Llama 4 Scout",
"description": "Open multimodal Llama model for long-context analysis and efficient agents",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2025-09-05",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 328000,
"input": 328000,
"output": 65536
},
"cost": {
"input": 0.085,
"output": 0.46,
"cache_read": 0.0425
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta-llama/llama-4-scout\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-4-scout\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-3.3-70b-instruct": {
"id": "meta-llama/llama-3.3-70b-instruct",
"name": "Llama 3.3 70b Instruct",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 16384
},
"cost": {
"input": 0.05,
"output": 0.23,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/meta-llama/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"abacusai/Dracarys-72B-Instruct": {
"id": "abacusai/Dracarys-72B-Instruct",
"name": "Llama 3.1 70B Dracarys 2",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-08-02",
"last_updated": "2025-08-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 8192
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/abacusai/Dracarys-72B-Instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"abacusai/Dracarys-72B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Gryphe/MythoMax-L2-13b": {
"id": "Gryphe/MythoMax-L2-13b",
"name": "MythoMax 13B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-08-08",
"last_updated": "2025-08-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 4096,
"input": 4096,
"output": 3686
},
"cost": {
"input": 0.1003,
"output": 0.1003,
"cache_read": 0.05015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Gryphe/MythoMax-L2-13b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Gryphe/MythoMax-L2-13b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o4-mini-high": {
"id": "openai/o4-mini-high",
"name": "OpenAI o4-mini high",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-mini",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2025-12-04",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/o4-mini-high\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/o4-mini-high\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-nano": {
"id": "openai/gpt-5-nano",
"name": "GPT 5 Nano",
"description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5-nano\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-mini-low": {
"id": "openai/o3-mini-low",
"name": "OpenAI o3-mini (Low)",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-mini",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-01-31",
"last_updated": "2025-01-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/o3-mini-low\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-mini-low\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-pro-2025-06-10": {
"id": "openai/o3-pro-2025-06-10",
"name": "OpenAI o3-pro (2025-06-10)",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2024-01-01",
"last_updated": "2025-06-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 100000
},
"cost": {
"input": 22,
"output": 88,
"cache_read": 11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/o3-pro-2025-06-10\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-pro-2025-06-10\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-nano": {
"id": "openai/gpt-4.1-nano",
"name": "GPT 4.1 Nano",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"input": 1047576,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-4.1-nano\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-pro": {
"id": "openai/gpt-5-pro",
"name": "GPT 5 Pro",
"description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 15,
"output": 120,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-mini-high": {
"id": "openai/o3-mini-high",
"name": "OpenAI o3-mini (High)",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-mini",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2024-01-01",
"last_updated": "2025-01-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/o3-mini-high\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-mini-high\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex-mini": {
"id": "openai/gpt-5.1-codex-mini",
"name": "GPT 5.1 Codex Mini",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.1-codex-mini\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-6-astra-pro": {
"id": "openai/gpt-6-astra-pro",
"name": "GPT 6 Astra Pro",
"description": "GPT 6 Astra in Pro reasoning mode. Uses additional model work for difficult tasks, with higher latency and token usage at the same per-token rates. Reasoning effort remains independently configurable.",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-6-astra-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-6-astra-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex": {
"id": "openai/gpt-5.1-codex",
"name": "GPT 5.1 Codex",
"description": "Codex GPT for repository edits, code review, and practical software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.1-codex\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-sol": {
"id": "openai/gpt-5.6-sol",
"name": "GPT 5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.6-sol\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-2024-08-06": {
"id": "openai/gpt-4o-2024-08-06",
"name": "GPT-4o (2024-08-06)",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-08-06",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-4o-2024-08-06\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-2024-08-06\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-codex": {
"id": "openai/gpt-5.2-codex",
"name": "GPT 5.2 Codex",
"description": "Code-specialist GPT for repository edits, reviews, and long-running software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.2-codex\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-2025-11-13": {
"id": "openai/gpt-5.1-2025-11-13",
"name": "GPT-5.1 (2025-11-13)",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2024-01-01",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.1-2025-11-13\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-2025-11-13\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-latest": {
"id": "openai/gpt-latest",
"name": "GPT Latest",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-29",
"last_updated": "2026-03-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-6-astra": {
"id": "openai/gpt-6-astra",
"name": "GPT 6 Astra",
"description": "GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.",
"family": "gpt-astra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-04-30",
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-6-astra\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-6-astra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-luna-pro": {
"id": "openai/gpt-5.6-luna-pro",
"name": "GPT 5.6 Luna Pro",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.6-luna-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-luna-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-terra-latest": {
"id": "openai/gpt-terra-latest",
"name": "GPT Terra Latest",
"description": "Compatibility alias that routes to GPT 5.6 Terra, the latest supported GPT Terra model.",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-11",
"last_updated": "2026-09-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-terra-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-terra-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-mini": {
"id": "openai/gpt-4.1-mini",
"name": "GPT 4.1 Mini",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"input": 1047576,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-4.1-mini\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4": {
"id": "openai/gpt-5.4",
"name": "GPT 5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.4\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 0.2,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-oss-20b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4-turbo": {
"id": "openai/gpt-4-turbo",
"name": "GPT-4 Turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 4096
},
"cost": {
"input": 10,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-4-turbo\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-sol-pro": {
"id": "openai/gpt-5.6-sol-pro",
"name": "GPT 5.6 Sol Pro",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.6-sol-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-sol-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-safeguard-20b": {
"id": "openai/gpt-oss-safeguard-20b",
"name": "GPT OSS Safeguard 20B",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-10-29",
"last_updated": "2025-10-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 0.075,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-oss-safeguard-20b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-safeguard-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1": {
"id": "openai/gpt-5.1",
"name": "GPT 5.1",
"description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex-max": {
"id": "openai/gpt-5.1-codex-max",
"name": "GPT 5.1 Codex Max",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 20,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.1-codex-max\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o1": {
"id": "openai/o1",
"name": "OpenAI o1",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": false,
"knowledge": "2023-09",
"release_date": "2024-12-05",
"last_updated": "2024-12-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 100000
},
"cost": {
"input": 15,
"output": 60,
"cache_read": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/o1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/o1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o": {
"id": "openai/gpt-4o",
"name": "GPT-4o",
"description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-4o\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-luna": {
"id": "openai/gpt-5.6-luna",
"name": "GPT 5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.6-luna\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-codex": {
"id": "openai/gpt-5.3-codex",
"name": "GPT 5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.3-codex\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini": {
"id": "openai/gpt-4o-mini",
"name": "GPT-4o mini",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-4o-mini\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o1-pro": {
"id": "openai/o1-pro",
"name": "OpenAI o1 Pro",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-pro",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": false,
"knowledge": "2023-09",
"release_date": "2025-03-19",
"last_updated": "2025-03-19",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 100000
},
"cost": {
"input": 150,
"output": 600,
"cache_read": 75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/o1-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/o1-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1": {
"id": "openai/gpt-4.1",
"name": "GPT 4.1",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"input": 1047576,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-4.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-nano": {
"id": "openai/gpt-5.4-nano",
"name": "GPT 5.4 Nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.4-nano\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-terra-pro": {
"id": "openai/gpt-5.6-terra-pro",
"name": "GPT 5.6 Terra Pro",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.6-terra-pro\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-terra-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-chat-latest": {
"id": "openai/gpt-chat-latest",
"name": "GPT Chat Latest",
"description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-03",
"last_updated": "2026-05-03",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-chat-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-chat-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-sol-latest": {
"id": "openai/gpt-sol-latest",
"name": "GPT Sol Latest",
"description": "Compatibility alias that routes to GPT 5.6 Sol, the latest supported GPT Sol model.",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-11",
"last_updated": "2026-09-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-sol-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-sol-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-luna-latest": {
"id": "openai/gpt-luna-latest",
"name": "GPT Luna Latest",
"description": "Compatibility alias that routes to GPT 5.6 Luna, the latest supported GPT Luna model.",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-11",
"last_updated": "2026-09-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-luna-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-luna-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-mini": {
"id": "openai/gpt-5.4-mini",
"name": "GPT 5.4 Mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.4-mini\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-3.5-turbo": {
"id": "openai/gpt-3.5-turbo",
"name": "GPT-3.5 Turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2021-09-01",
"release_date": "2023-03-01",
"last_updated": "2023-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16385,
"input": 16385,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-3.5-turbo\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-3.5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-astra-latest": {
"id": "openai/gpt-astra-latest",
"name": "GPT Astra Latest",
"description": "Compatibility alias that routes to GPT 6 Astra, the latest supported GPT Astra model.",
"family": "gpt-astra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-11",
"last_updated": "2026-09-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-astra-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-astra-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-mini": {
"id": "openai/gpt-5-mini",
"name": "GPT 5 Mini",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5-mini\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 0.35,
"output": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-oss-120b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-terra": {
"id": "openai/gpt-5.6-terra",
"name": "GPT 5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.6-terra\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2": {
"id": "openai/gpt-5.2",
"name": "GPT 5.2",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5": {
"id": "openai/gpt-5",
"name": "GPT 5",
"description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o4-mini": {
"id": "openai/o4-mini",
"name": "OpenAI o4-mini",
"description": "Fast o-series model for compact reasoning, coding, and tool use",
"family": "o-mini",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/o4-mini\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/o4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-mini": {
"id": "openai/o3-mini",
"name": "OpenAI o3-mini",
"description": "Smaller o-series reasoner for economical coding, math, and planning tasks",
"family": "o-mini",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2024-12-20",
"last_updated": "2025-01-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/o3-mini\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3": {
"id": "openai/o3",
"name": "OpenAI o3",
"description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
"family": "o",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 200000,
"output": 100000
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/o3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT 5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 1050000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-5.5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-2024-11-20": {
"id": "openai/gpt-4o-2024-11-20",
"name": "GPT-4o (2024-11-20)",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-11-20",
"last_updated": "2024-11-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/openai/gpt-4o-2024-11-20\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-2024-11-20\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-latest": {
"id": "moonshotai/kimi-latest",
"name": "Kimi Latest",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-03",
"last_updated": "2026-05-03",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 943718
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/kimi-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.7-code-highspeed": {
"id": "moonshotai/kimi-k2.7-code-highspeed",
"name": "Kimi K2.7 Code High-Speed",
"description": "Lower-latency Kimi Code variant for interactive edits and coding-agent loops",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 1.9,
"output": 8,
"cache_read": 0.32
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/kimi-k2.7-code-highspeed\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.7-code-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.6": {
"id": "moonshotai/kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 2.6,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/kimi-k2.6\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2-Instruct-0905": {
"id": "moonshotai/Kimi-K2-Instruct-0905",
"name": "Kimi K2 0905",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 100352
},
"cost": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/Kimi-K2-Instruct-0905\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2-Instruct-0905\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.7-code": {
"id": "moonshotai/kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 65536
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/kimi-k2.7-code\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-thinking": {
"id": "moonshotai/kimi-k2-thinking",
"name": "Kimi K2 Thinking",
"description": "Thinking Kimi model for slower research passes, planning, and hard technical questions",
"family": "kimi-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-11-06",
"last_updated": "2025-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 98304
},
"cost": {
"input": 0.6,
"output": 2.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/kimi-k2-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k3": {
"id": "moonshotai/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 943718
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/kimi-k3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-instruct": {
"id": "moonshotai/kimi-k2-instruct",
"name": "Kimi K2 Instruct",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2024-01-01",
"last_updated": "2025-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 8192
},
"cost": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/kimi-k2-instruct\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.5": {
"id": "moonshotai/kimi-k2.5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.9,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/kimi-k2.5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.6:thinking": {
"id": "moonshotai/kimi-k2.6:thinking",
"name": "Kimi K2.6 Thinking",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 2.6,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/kimi-k2.6:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.6:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-instruct-0711": {
"id": "moonshotai/kimi-k2-instruct-0711",
"name": "Kimi K2 0711",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2024-01-01",
"last_updated": "2025-07-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 8192
},
"cost": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/kimi-k2-instruct-0711\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-instruct-0711\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.5:thinking": {
"id": "moonshotai/kimi-k2.5:thinking",
"name": "Kimi K2.5 Thinking",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.9,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/moonshotai/kimi-k2.5:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.5:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cohere/command-r-plus-08-2024": {
"id": "cohere/command-r-plus-08-2024",
"name": "Cohere: Command R+",
"description": "Cohere's RAG workhorse for long-context enterprise search and tool use",
"family": "command-r",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-06-01",
"release_date": "2024-08-30",
"last_updated": "2024-08-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 4096
},
"cost": {
"input": 2.856,
"output": 14.246,
"cache_read": 1.428
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/cohere/command-r-plus-08-2024\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"cohere/command-r-plus-08-2024\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"upstage/solar-pro-3": {
"id": "upstage/solar-pro-3",
"name": "Solar Pro 3",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "solar-pro",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"input": 131072,
"output": 117964
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/upstage/solar-pro-3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"upstage/solar-pro-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"upstage/solar-pro4": {
"id": "upstage/solar-pro4",
"name": "Solar Pro 4",
"description": "Upstage's flagship model, specialized for agentic use",
"family": "solar-pro",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02",
"release_date": "2026-08-06",
"last_updated": "2026-08-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 524288,
"input": 524288,
"output": 131072
},
"cost": {
"input": 0.03,
"output": 0.12,
"cache_read": 0.006
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/upstage/solar-pro4\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"upstage/solar-pro4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"upstage/solar-pro4:thinking": {
"id": "upstage/solar-pro4:thinking",
"name": "Solar Pro 4 Thinking",
"description": "Upstage's flagship model, specialized for agentic use",
"family": "solar-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02",
"release_date": "2026-08-06",
"last_updated": "2026-08-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 524288,
"input": 524288,
"output": 131072
},
"cost": {
"input": 0.03,
"output": 0.12,
"cache_read": 0.006
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/upstage/solar-pro4:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"upstage/solar-pro4:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"arcee-ai/trinity-large-thinking": {
"id": "arcee-ai/trinity-large-thinking",
"name": "Trinity Large Thinking",
"description": "Reasoning-optimized 398B MoE agent model with extended thinking for long-horizon and multi-turn tool use",
"family": "trinity",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 80000
},
"cost": {
"input": 0.25,
"output": 0.9,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/arcee-ai/trinity-large-thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"arcee-ai/trinity-large-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/hy3": {
"id": "tencent/hy3",
"name": "Tencent Hy3",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 128000
},
"cost": {
"input": 0.066,
"output": 0.26,
"cache_read": 0.029
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/tencent/hy3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"tencent/hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/hy4-preview": {
"id": "tencent/hy4-preview",
"name": "Tencent Hy4 Preview",
"description": "A next-generation productivity model with significantly enhanced Agent and complex task execution capabilities.",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-28",
"last_updated": "2026-08-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 64000
},
"cost": {
"input": 0.834,
"output": 2.501,
"cache_read": 0.042
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/tencent/hy4-preview\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"tencent/hy4-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TheDrummer/skyfall-36b-v2": {
"id": "TheDrummer/skyfall-36b-v2",
"name": "TheDrummer Skyfall 36B V2",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-03-10",
"last_updated": "2025-03-10",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 29491
},
"cost": {
"input": 0.55,
"output": 0.8,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TheDrummer/skyfall-36b-v2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TheDrummer/skyfall-36b-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TheDrummer/UnslopNemo-12B-v4.1": {
"id": "TheDrummer/UnslopNemo-12B-v4.1",
"name": "UnslopNemo 12b v4",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"input": 8192,
"output": 26214
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TheDrummer/UnslopNemo-12B-v4.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TheDrummer/UnslopNemo-12B-v4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TheDrummer/Cydonia-24B-v4.3": {
"id": "TheDrummer/Cydonia-24B-v4.3",
"name": "The Drummer Cydonia 24B v4.3",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-25",
"last_updated": "2025-12-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.15,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TheDrummer/Cydonia-24B-v4.3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TheDrummer/Cydonia-24B-v4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TheDrummer/Artemis-v1.1": {
"id": "TheDrummer/Artemis-v1.1",
"name": "TheDrummer/Artemis v1.1",
"description": "TheDrummer's Artemis v1.1 is a Gemma 4 31B fine-tune for creative writing, expressive dialogue, and roleplay, with optional thinking and a 262K context window.",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-09-06",
"last_updated": "2026-09-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.45,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TheDrummer/Artemis-v1.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TheDrummer/Artemis-v1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TheDrummer/Cydonia-24B-v2": {
"id": "TheDrummer/Cydonia-24B-v2",
"name": "The Drummer Cydonia 24B v2",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-02-17",
"last_updated": "2025-02-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 0.1003,
"output": 0.1207,
"cache_read": 0.05015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TheDrummer/Cydonia-24B-v2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TheDrummer/Cydonia-24B-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TheDrummer/Cydonia-24B-v4": {
"id": "TheDrummer/Cydonia-24B-v4",
"name": "The Drummer Cydonia 24B v4",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-22",
"last_updated": "2025-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 0.2006,
"output": 0.2414,
"cache_read": 0.1003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TheDrummer/Cydonia-24B-v4\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TheDrummer/Cydonia-24B-v4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TheDrummer/Anubis-70B-v1.1": {
"id": "TheDrummer/Anubis-70B-v1.1",
"name": "Anubis 70B v1.1",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"input": 32000,
"output": 16384
},
"cost": {
"input": 0.31,
"output": 0.31,
"cache_read": 0.155
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TheDrummer/Anubis-70B-v1.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TheDrummer/Anubis-70B-v1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TheDrummer/Magidonia-24B-v4.3": {
"id": "TheDrummer/Magidonia-24B-v4.3",
"name": "The Drummer Magidonia 24B v4.3",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-25",
"last_updated": "2025-12-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 0.1003,
"output": 0.1207,
"cache_read": 0.05015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TheDrummer/Magidonia-24B-v4.3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TheDrummer/Magidonia-24B-v4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TheDrummer/Cydonia-24B-v4.1": {
"id": "TheDrummer/Cydonia-24B-v4.1",
"name": "The Drummer Cydonia 24B v4.1",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-08-19",
"last_updated": "2025-08-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 131072,
"output": 117964
},
"cost": {
"input": 0.35,
"output": 0.55,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TheDrummer/Cydonia-24B-v4.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TheDrummer/Cydonia-24B-v4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TheDrummer/Anubis-70B-v1": {
"id": "TheDrummer/Anubis-70B-v1",
"name": "Anubis 70B v1",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"input": 65536,
"output": 16384
},
"cost": {
"input": 0.31,
"output": 0.31,
"cache_read": 0.155
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TheDrummer/Anubis-70B-v1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TheDrummer/Anubis-70B-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"TheDrummer/Rocinante-12B-v1.1": {
"id": "TheDrummer/Rocinante-12B-v1.1",
"name": "Rocinante 12b",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 0.408,
"output": 0.595,
"cache_read": 0.204
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/TheDrummer/Rocinante-12B-v1.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"TheDrummer/Rocinante-12B-v1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"soob3123/Veiled-Calla-12B": {
"id": "soob3123/Veiled-Calla-12B",
"name": "Veiled Calla 12B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-04-13",
"last_updated": "2025-04-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 8192
},
"cost": {
"input": 0.3,
"output": 0.3,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/soob3123/Veiled-Calla-12B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"soob3123/Veiled-Calla-12B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"soob3123/amoral-gemma3-27B-v2": {
"id": "soob3123/amoral-gemma3-27B-v2",
"name": "Amoral Gemma3 27B v2",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-05-23",
"last_updated": "2025-05-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 8192
},
"cost": {
"input": 0.3,
"output": 0.3,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/soob3123/amoral-gemma3-27B-v2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"soob3123/amoral-gemma3-27B-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"soob3123/GrayLine-Qwen3-8B": {
"id": "soob3123/GrayLine-Qwen3-8B",
"name": "Grayline Qwen3 8B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 0.3,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/soob3123/GrayLine-Qwen3-8B\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"soob3123/GrayLine-Qwen3-8B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-medium-3.5:thinking": {
"id": "mistral/mistral-medium-3.5:thinking",
"name": "Mistral Medium 3.5 Thinking",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-medium",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-04-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 32768
},
"cost": {
"input": 1.5,
"output": 7.5,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistral/mistral-medium-3.5:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-medium-3.5:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-medium-3.5": {
"id": "mistral/mistral-medium-3.5",
"name": "Mistral Medium 3.5",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-medium",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-04-29",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 32768
},
"cost": {
"input": 1.5,
"output": 7.5,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/mistral/mistral-medium-3.5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-medium-3.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nanogpt/coding-router:low": {
"id": "nanogpt/coding-router:low",
"name": "Coding Router Low",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-12",
"last_updated": "2026-05-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nanogpt/coding-router:low\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nanogpt/coding-router:low\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nanogpt/coding-router:high": {
"id": "nanogpt/coding-router:high",
"name": "Coding Router High",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-12",
"last_updated": "2026-05-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 1.1,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nanogpt/coding-router:high\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nanogpt/coding-router:high\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nanogpt/coding-router": {
"id": "nanogpt/coding-router",
"name": "Coding Router",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-12",
"last_updated": "2026-05-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 1.1,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nanogpt/coding-router\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nanogpt/coding-router\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nanogpt/coding-router:max": {
"id": "nanogpt/coding-router:max",
"name": "Coding Router Max",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-12",
"last_updated": "2026-05-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nanogpt/coding-router:max\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nanogpt/coding-router:max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nanogpt/coding-router:medium": {
"id": "nanogpt/coding-router:medium",
"name": "Coding Router Medium",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-12",
"last_updated": "2026-05-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 128000
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/nanogpt/coding-router:medium\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"nanogpt/coding-router:medium\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"liquid/lfm-2.5-2.6b": {
"id": "liquid/lfm-2.5-2.6b",
"name": "LFM2.5 2.6B",
"description": "Liquid AI's compact 2.6B reasoning model for agent workflows, data extraction, RAG, and long-context processing. It supports tool calling and structured output, but Liquid advises against using it for agentic coding. Warning: prompts and responses may be logged and used for model training or service improvement; do not send sensitive data.",
"family": "liquid",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/liquid/lfm-2.5-2.6b\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"liquid/lfm-2.5-2.6b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7": {
"id": "z-ai/glm-4.7",
"name": "GLM 4.7",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 65535
},
"cost": {
"input": 0.2,
"output": 0.8,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.7\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.5v:thinking": {
"id": "z-ai/glm-4.5v:thinking",
"name": "GLM 4.5V Thinking",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-08-11",
"last_updated": "2025-08-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"input": 65536,
"output": 16384
},
"cost": {
"input": 0.6,
"output": 1.8,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.5v:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.5v:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.6": {
"id": "z-ai/glm-4.6",
"name": "GLM 4.6",
"description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 65535
},
"cost": {
"input": 0.35,
"output": 1.4,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.6\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.6v": {
"id": "z-ai/glm-4.6v",
"name": "GLM 4.6V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 24000
},
"cost": {
"input": 0.3,
"output": 0.9,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.6v\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.6v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.6v-original": {
"id": "z-ai/glm-4.6v-original",
"name": "GLM 4.6V Original",
"description": "GLM-4.6V scales its context window to 128k tokens in training, and achieves SoTA performance in visual understanding among models of similar parameter scales. Integrates native Function Calling capabilities, bridging 'visual perception' and 'executable action' for multimodal agents. Direct via Z-AI (Zhipu).",
"family": "glm",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 24000
},
"cost": {
"input": 0.6,
"output": 0.9,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.6v-original\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.6v-original\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.3:thinking": {
"id": "z-ai/glm-5.3:thinking",
"name": "GLM 5.3 Thinking",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5.3:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/GLM-4.6-turbo": {
"id": "z-ai/GLM-4.6-turbo",
"name": "GLM 4.6 Turbo",
"description": "Fast variant of GLM 4.6 for general chat, coding, and analysis with improved latency and strong reasoning.",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-10-02",
"last_updated": "2025-10-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"input": 204800,
"output": 131072
},
"cost": {
"input": 1,
"output": 3,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/GLM-4.6-turbo\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/GLM-4.6-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/GLM-4.5-Air:thinking": {
"id": "z-ai/GLM-4.5-Air:thinking",
"name": "GLM 4.5 Air (Thinking)",
"description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
"family": "glm-air",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 98304
},
"cost": {
"input": 0.12,
"output": 0.8,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/GLM-4.5-Air:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/GLM-4.5-Air:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.6:thinking": {
"id": "z-ai/glm-4.6:thinking",
"name": "GLM 4.6 Thinking",
"description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 65535
},
"cost": {
"input": 0.35,
"output": 1.4,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.6:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.6:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.2": {
"id": "z-ai/glm-5.2",
"name": "GLM 5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 0.42,
"output": 1.32,
"cache_read": 0.078
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5.2\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7-original": {
"id": "z-ai/glm-4.7-original",
"name": "GLM 4.7 Original",
"description": "GLM-4.7 is a next-gen GLM series text model with stronger reasoning, long-context chat, and reliable tool use. Routed directly via Z-AI (Zhipu).",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 65535
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.7-original\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7-original\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7-flash:thinking": {
"id": "z-ai/glm-4.7-flash:thinking",
"name": "GLM 4.7 Flash Thinking",
"description": "Budget GLM lane for fast coding help, routing, and everyday automation",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 128000
},
"cost": {
"input": 0.07,
"output": 0.4,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.7-flash:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7-flash:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7:thinking": {
"id": "z-ai/glm-4.7:thinking",
"name": "GLM 4.7 Thinking",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 65535
},
"cost": {
"input": 0.2,
"output": 0.8,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.7:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5v-turbo:thinking": {
"id": "z-ai/glm-5v-turbo:thinking",
"name": "GLM 5V Turbo Thinking",
"description": "Fast GLM vision model for screenshots, documents, and multimodal agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-04-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 202800,
"input": 202800,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5v-turbo:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5v-turbo:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7-flash-original": {
"id": "z-ai/glm-4.7-flash-original",
"name": "GLM 4.7 Flash Original",
"description": "GLM-4.7-Flash is a lightweight 30B model optimized for coding and agentic tasks. Balances high performance with efficiency, perfect for local deployment.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 128000
},
"cost": {
"input": 0.07,
"output": 0.4,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.7-flash-original\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7-flash-original\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/GLM-4.5:thinking": {
"id": "z-ai/GLM-4.5:thinking",
"name": "GLM 4.5 (Thinking)",
"description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.3,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/GLM-4.5:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/GLM-4.5:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5-original:thinking": {
"id": "z-ai/glm-5-original:thinking",
"name": "GLM 5 Original Thinking",
"description": "GLM-5 original with extended thinking capabilities for complex reasoning.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-02-11",
"last_updated": "2026-02-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 128000
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5-original:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5-original:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.3-flash": {
"id": "z-ai/glm-5.3-flash",
"name": "GLM 5.3 Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 0.075,
"output": 0.25,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5.3-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.5": {
"id": "z-ai/glm-4.5",
"name": "GLM 4.5",
"description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.3,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/GLM-4.5-Air": {
"id": "z-ai/GLM-4.5-Air",
"name": "GLM 4.5 Air",
"description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
"family": "glm-air",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 98304
},
"cost": {
"input": 0.12,
"output": 0.8,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/GLM-4.5-Air\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/GLM-4.5-Air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7-original:thinking": {
"id": "z-ai/glm-4.7-original:thinking",
"name": "GLM 4.7 Original Thinking",
"description": "GLM-4.7 original with extended thinking capabilities for complex reasoning.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 65535
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.7-original:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7-original:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.5v": {
"id": "z-ai/glm-4.5v",
"name": "GLM 4.5V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-08-11",
"last_updated": "2025-08-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"input": 65536,
"output": 16384
},
"cost": {
"input": 0.6,
"output": 1.8,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.5v\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.5v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.1:thinking": {
"id": "z-ai/glm-5.1:thinking",
"name": "GLM 5.1 Thinking",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 131072
},
"cost": {
"input": 0.75,
"output": 2.6,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5.1:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.1:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5": {
"id": "z-ai/glm-5",
"name": "GLM 5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 128000
},
"cost": {
"input": 0.5,
"output": 2.55,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/GLM-4.6-turbo:thinking": {
"id": "z-ai/GLM-4.6-turbo:thinking",
"name": "GLM 4.6 Turbo (Thinking)",
"description": "GLM 4.6 Turbo with thinking mode enabled for enhanced reasoning; shows internal reasoning and supports long context.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"release_date": "2025-10-02",
"last_updated": "2025-10-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"input": 204800,
"output": 131072
},
"cost": {
"input": 1,
"output": 3,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/GLM-4.6-turbo:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/GLM-4.6-turbo:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.1": {
"id": "z-ai/glm-5.1",
"name": "GLM 5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 131072
},
"cost": {
"input": 0.75,
"output": 2.6,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5.1\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.2:thinking": {
"id": "z-ai/glm-5.2:thinking",
"name": "GLM 5.2 Thinking",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 0.42,
"output": 1.32,
"cache_read": 0.078
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5.2:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.2:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-latest": {
"id": "z-ai/glm-latest",
"name": "GLM Latest",
"description": "Compatibility alias that routes to the newest thinking GLM model. Currently routes to GLM 5.2 Thinking.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-03",
"last_updated": "2026-05-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-latest\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5-turbo": {
"id": "z-ai/glm-5-turbo",
"name": "GLM 5 Turbo",
"description": "Faster GLM-5 lane for coding agents that need lower latency",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 202800,
"input": 202800,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5-turbo\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7-flash-original:thinking": {
"id": "z-ai/glm-4.7-flash-original:thinking",
"name": "GLM 4.7 Flash Original Thinking",
"description": "GLM-4.7-Flash with extended thinking capabilities for complex reasoning. Lightweight 30B model optimized for coding and agentic tasks.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 128000
},
"cost": {
"input": 0.07,
"output": 0.4,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.7-flash-original:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7-flash-original:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5-original": {
"id": "z-ai/glm-5-original",
"name": "GLM 5 Original",
"description": "GLM-5 is Zhipu's latest flagship model with advanced reasoning and instruction following. Routed directly via Z-AI (Zhipu).",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-02-11",
"last_updated": "2026-02-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 128000
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5-original\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5-original\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.3": {
"id": "z-ai/glm-5.3",
"name": "GLM 5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5.3\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5v-turbo": {
"id": "z-ai/glm-5v-turbo",
"name": "GLM 5V Turbo",
"description": "Fast GLM vision model for screenshots, documents, and multimodal agent tasks",
"family": "glm",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-04-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 202800,
"input": 202800,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5v-turbo\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5v-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.3-flash-uncensored": {
"id": "z-ai/glm-5.3-flash-uncensored",
"name": "GLM 5.3 Flash Uncensored",
"description": "GLM 5.3 Flash Uncensored is an uncensored fine-tune of the efficient 320B mixture-of-experts reasoning model, built for unrestricted chat, creative writing, coding, agentic work, tool use, and long-context tasks.",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-29",
"last_updated": "2026-08-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"input": 1048576,
"output": 32768
},
"cost": {
"input": 0.35,
"output": 1.4,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5.3-flash-uncensored\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3-flash-uncensored\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.6-original": {
"id": "z-ai/glm-4.6-original",
"name": "GLM 4.6 Original",
"description": "GLM-4.6, Zhipu's flagship text model with 256K context window and advanced reasoning capabilities. Direct via Z-AI (Zhipu).",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 65535
},
"cost": {
"input": 0.35,
"output": 1.4,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.6-original\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.6-original\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5:thinking": {
"id": "z-ai/glm-5:thinking",
"name": "GLM 5 Thinking",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 128000
},
"cost": {
"input": 0.5,
"output": 2.55,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-5:thinking\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5:thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7-flash": {
"id": "z-ai/glm-4.7-flash",
"name": "GLM 4.7 Flash",
"description": "Budget GLM lane for fast coding help, routing, and everyday automation",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 128000
},
"cost": {
"input": 0.07,
"output": 0.4,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/z-ai/glm-4.7-flash\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Doctor-Shotgun/MS3.2-24B-Magnum-Diamond": {
"id": "Doctor-Shotgun/MS3.2-24B-Magnum-Diamond",
"name": "MS3.2 24B Magnum Diamond",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 0.493,
"output": 0.493,
"cache_read": 0.2465
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/Doctor-Shotgun/MS3.2-24B-Magnum-Diamond\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"Doctor-Shotgun/MS3.2-24B-Magnum-Diamond\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"THUDM/GLM-4-9B-0414": {
"id": "THUDM/GLM-4-9B-0414",
"name": "GLM 4 9B 0414",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"input": 32000,
"output": 8000
},
"cost": {
"input": 0.2,
"output": 0.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/THUDM/GLM-4-9B-0414\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"THUDM/GLM-4-9B-0414\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"THUDM/GLM-Z1-9B-0414": {
"id": "THUDM/GLM-Z1-9B-0414",
"name": "GLM Z1 9B 0414",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm-z",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"input": 32000,
"output": 8000
},
"cost": {
"input": 0.2,
"output": 0.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/THUDM/GLM-Z1-9B-0414\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"THUDM/GLM-Z1-9B-0414\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"THUDM/GLM-4-32B-0414": {
"id": "THUDM/GLM-4-32B-0414",
"name": "GLM 4 32B 0414",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2024-01-01",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 128000,
"output": 65536
},
"cost": {
"input": 0.2,
"output": 0.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nano-gpt/THUDM/GLM-4-32B-0414\", apiKey: processEnvironment[\"NANO_GPT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://nano-gpt.com/api/v1\")!,\n apiKey: processEnvironment[\"NANO_GPT_API_KEY\"]\n)\nlet session = provider.model(\"THUDM/GLM-4-32B-0414\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"watsonx": {
"id": "watsonx",
"name": "watsonx.ai",
"baseURL": "",
"npm": "watsonx-ai-provider",
"swiftDriver": "openaiChat",
"env": [
"WATSONX_AI_APIKEY",
"WATSONX_AI_PROJECT_ID"
],
"doc": "https://www.ibm.com/docs/en/watsonx/saas?topic=solutions-supported-foundation-models",
"modelCount": 5,
"models": {
"mistralai/mistral-small-3-1-24b-instruct-2503": {
"id": "mistralai/mistral-small-3-1-24b-instruct-2503",
"name": "Mistral Small 3.1 24B",
"description": "Efficient multimodal model for instruction following, coding, reasoning, and function calling",
"family": "mistral-small",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2025-03-17",
"last_updated": "2025-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.106,
"output": 0.318
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"watsonx/mistralai/mistral-small-3-1-24b-instruct-2503\", apiKey: processEnvironment[\"WATSONX_AI_APIKEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"WATSONX_AI_APIKEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-small-3-1-24b-instruct-2503\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ibm/granite-4-h-small": {
"id": "ibm/granite-4-h-small",
"name": "Granite-4.0-H-Small",
"description": "Open-weight hybrid model for enterprise chat, coding, retrieval-augmented generation, and tool-calling workloads",
"family": "granite",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-02",
"last_updated": "2025-10-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.0636,
"output": 0.265
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"watsonx/ibm/granite-4-h-small\", apiKey: processEnvironment[\"WATSONX_AI_APIKEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"WATSONX_AI_APIKEY\"]\n)\nlet session = provider.model(\"ibm/granite-4-h-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-4-maverick-17b-128e-instruct-fp8": {
"id": "meta-llama/llama-4-maverick-17b-128e-instruct-fp8",
"name": "Llama 4 Maverick 17B 128E Instruct FP8",
"description": "Open multimodal Llama for strong reasoning with efficient everyday serving",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.371,
"output": 1.484
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"watsonx/meta-llama/llama-4-maverick-17b-128e-instruct-fp8\", apiKey: processEnvironment[\"WATSONX_AI_APIKEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"WATSONX_AI_APIKEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-4-maverick-17b-128e-instruct-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-3-3-70b-instruct": {
"id": "meta-llama/llama-3-3-70b-instruct",
"name": "Llama-3.3-70B-Instruct",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 4096
},
"cost": {
"input": 0.7526,
"output": 0.7526
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"watsonx/meta-llama/llama-3-3-70b-instruct\", apiKey: processEnvironment[\"WATSONX_AI_APIKEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"WATSONX_AI_APIKEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-3-3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.159,
"output": 0.636
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"watsonx/openai/gpt-oss-120b\", apiKey: processEnvironment[\"WATSONX_AI_APIKEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"WATSONX_AI_APIKEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"digitalocean": {
"id": "digitalocean",
"name": "DigitalOcean",
"baseURL": "https://inference.do-ai.run/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"DIGITALOCEAN_ACCESS_TOKEN"
],
"doc": "https://docs.digitalocean.com/products/gradient-ai-platform/details/models/",
"modelCount": 96,
"models": {
"openai-gpt-4o": {
"id": "openai-gpt-4o",
"name": "OpenAI GPT-4o",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-4o\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.2-pro": {
"id": "openai-gpt-5.2-pro",
"name": "OpenAI GPT-5.2 Pro",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 21,
"output": 168
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.2-pro\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bge-reranker-v2-m3": {
"id": "bge-reranker-v2-m3",
"name": "BGE Reranker v2 M3",
"description": "Reranking model for improving retrieval quality in search and recommendation systems",
"family": "bge",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-03-12",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 1
},
"cost": {
"input": 0.01,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/bge-reranker-v2-m3\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"bge-reranker-v2-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-opus-4.6": {
"id": "anthropic-claude-opus-4.6",
"name": "Anthropic Claude Opus 4.6",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-opus-4.6\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-opus-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-o3": {
"id": "openai-o3",
"name": "OpenAI o3",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-o3\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro-0813": {
"id": "deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.32,
"output": 3.96,
"cache_read": 0.044
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.08,
"output": 0.252,
"cache_read": 0.0252
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3": {
"id": "deepseek-v3",
"name": "DeepSeek V3",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-12-26",
"last_updated": "2025-03-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/deepseek-v3\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-2.5-14b-instruct": {
"id": "qwen-2.5-14b-instruct",
"name": "Qwen 2.5 14B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2024-09-19",
"last_updated": "2024-09-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/qwen-2.5-14b-instruct\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"qwen-2.5-14b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia-nemotron-3-super-120b": {
"id": "nvidia-nemotron-3-super-120b",
"name": "NVIDIA Nemotron 3 Super 120B (Public Preview)",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 0.65,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/nvidia-nemotron-3-super-120b\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"nvidia-nemotron-3-super-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-flash": {
"id": "qwen3-coder-flash",
"name": "Qwen3 Coder Flash",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.45,
"output": 1.7,
"cache_read": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/qwen3-coder-flash\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"qwen3-coder-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.6-terra": {
"id": "openai-gpt-5.6-terra",
"name": "OpenAI GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.6-terra\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31B-it": {
"id": "gemma-4-31B-it",
"name": "Gemma 4",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 8192
},
"cost": {
"input": 0.18,
"output": 0.5,
"cache_read": 0.036
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/gemma-4-31B-it\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"gemma-4-31B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba-qwen3-32b": {
"id": "alibaba-qwen3-32b",
"name": "Qwen3 32B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-30",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.25,
"output": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/alibaba-qwen3-32b\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"alibaba-qwen3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-image-1.5": {
"id": "openai-gpt-image-1.5",
"name": "OpenAI GPT Image 1.5",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-image",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-11-25",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image",
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 16384
},
"cost": {
"input": 5,
"output": 10,
"cache_read": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-image-1.5\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-image-1.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/kimi-k2.6\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-opus-4": {
"id": "anthropic-claude-opus-4",
"name": "Claude Opus 4",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-opus-4\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-opus-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-6-astra": {
"id": "openai-gpt-6-astra",
"name": "OpenAI GPT-6 Astra",
"description": "GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.",
"family": "gpt-astra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-04-30",
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"tiers": [
{
"input": 20,
"output": 75,
"cache_read": 2,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 20,
"output": 75,
"cache_read": 2
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-6-astra\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-6-astra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.7,
"output": 2.2,
"cache_read": 0.105
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/glm-5.2\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"arcee-trinity-large-thinking": {
"id": "arcee-trinity-large-thinking",
"name": "Arcee Trinity Large Thinking (Public Preview)",
"description": "Reasoning-optimized 398B MoE agent model with extended thinking for long-horizon and multi-turn tool use",
"family": "trinity",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32000
},
"cost": {
"input": 0.25,
"output": 0.9,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/arcee-trinity-large-thinking\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"arcee-trinity-large-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-opus-4.7": {
"id": "anthropic-claude-opus-4.7",
"name": "Anthropic Claude Opus 4.7",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-opus-4.7\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-opus-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-fable-5": {
"id": "anthropic-claude-fable-5",
"name": "Anthropic Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-06-09",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-fable-5\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-3.5-sonnet": {
"id": "anthropic-claude-3.5-sonnet",
"name": "Claude 3.5 Sonnet",
"description": "Legacy model retained for compatibility with older integrations",
"family": "claude-sonnet",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-06-20",
"last_updated": "2024-10-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"status": "deprecated",
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-3.5-sonnet\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-3.5-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.5": {
"id": "minimax-m2.5",
"name": "MiniMax M2.5 (Public Preview)",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax-m2.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08",
"release_date": "2026-02-12",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/minimax-m2.5\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-4.1": {
"id": "openai-gpt-4.1",
"name": "OpenAI GPT-4.1",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-4.1\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.3-codex": {
"id": "openai-gpt-5.3-codex",
"name": "OpenAI GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.3-codex\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-oss-120b": {
"id": "openai-gpt-oss-120b",
"name": "OpenAI GPT-oss-120b",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2025-08-05",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.055,
"output": 0.385,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-oss-120b\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.2": {
"id": "openai-gpt-5.2",
"name": "OpenAI GPT-5.2",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.2\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-4.5-haiku": {
"id": "anthropic-claude-4.5-haiku",
"name": "Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-4.5-haiku\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-4.5-haiku\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"e5-large-v2": {
"id": "e5-large-v2",
"name": "E5 Large v2",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2023-05-19",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512,
"output": 1024
},
"cost": {
"input": 0.02,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/e5-large-v2\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"e5-large-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.6-luna": {
"id": "openai-gpt-5.6-luna",
"name": "OpenAI GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"tiers": [
{
"input": 0.4,
"output": 1.8,
"cache_read": 0.04,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.04
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.6-luna\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-5-sonnet": {
"id": "anthropic-claude-5-sonnet",
"name": "Anthropic Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-5-sonnet\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-5-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-oss-20b": {
"id": "openai-gpt-oss-20b",
"name": "OpenAI GPT-oss-20b",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2025-08-05",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.05,
"output": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-oss-20b\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-3.2": {
"id": "deepseek-3.2",
"name": "Deepseek 3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-12-02",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 0.25,
"output": 0.8,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/deepseek-3.2\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"multi-qa-mpnet-base-dot-v1": {
"id": "multi-qa-mpnet-base-dot-v1",
"name": "Multi-QA-mpnet-base-dot-v1",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2021-08-30",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512,
"output": 768
},
"cost": {
"input": 0.009,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/multi-qa-mpnet-base-dot-v1\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"multi-qa-mpnet-base-dot-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-3.7-sonnet": {
"id": "anthropic-claude-3.7-sonnet",
"name": "Claude 3.7 Sonnet",
"description": "Legacy model retained for compatibility with older integrations",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-11",
"release_date": "2025-02-24",
"last_updated": "2025-02-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"status": "deprecated",
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-3.7-sonnet\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-3.7-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-nano-30b": {
"id": "nemotron-3-nano-30b",
"name": "Nemotron 3 Nano 30B A3B",
"description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/nemotron-3-nano-30b\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"nemotron-3-nano-30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gte-large-en-v1.5": {
"id": "gte-large-en-v1.5",
"name": "GTE Large (v1.5)",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-03-27",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 1024
},
"cost": {
"input": 0.09,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/gte-large-en-v1.5\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"gte-large-en-v1.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-397b-a17b": {
"id": "qwen3.5-397b-a17b",
"name": "Qwen 3.5 397B A17B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.55,
"output": 3.5,
"cache_read": 0.111
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5": {
"id": "openai-gpt-5",
"name": "OpenAI GPT-5",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-4-maverick": {
"id": "llama-4-maverick",
"name": "Llama 4 Maverick",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.2,
"output": 0.696
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/llama-4-maverick\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"llama-4-maverick\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama3.3-70b-instruct": {
"id": "llama3.3-70b-instruct",
"name": "Llama 3.3 Instruct (70B)",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.65,
"output": 0.65
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/llama3.3-70b-instruct\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"llama3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"all-mini-lm-l6-v2": {
"id": "all-mini-lm-l6-v2",
"name": "All-MiniLM-L6-v2",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2021-08-30",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256,
"output": 384
},
"cost": {
"input": 0.009,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/all-mini-lm-l6-v2\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"all-mini-lm-l6-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-sonnet-4": {
"id": "anthropic-claude-sonnet-4",
"name": "Claude Sonnet 4",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75,
"tiers": [
{
"input": 6,
"output": 22.5,
"cache_read": 0.3,
"cache_write": 3.75,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 6,
"output": 22.5,
"cache_read": 0.3,
"cache_write": 3.75
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-sonnet-4\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-sonnet-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 2.55,
"output": 12.95,
"cache_read": 0.285
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/kimi-k3\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.4-mini": {
"id": "openai-gpt-5.4-mini",
"name": "OpenAI GPT-5.4 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.4-mini\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-opus-4.8": {
"id": "anthropic-claude-opus-4.8",
"name": "Anthropic Claude Opus 4.8",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-28",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-opus-4.8\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-opus-4.8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.4-pro": {
"id": "openai-gpt-5.4-pro",
"name": "OpenAI GPT-5.4 Pro",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180,
"tiers": [
{
"input": 60,
"output": 270,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 60,
"output": 270
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.4-pro\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-4-flash": {
"id": "deepseek-4-flash",
"name": "Deepseek V4 Flash",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2026-05-27",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.0679,
"output": 0.168,
"cache_read": 0.0168
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/deepseek-4-flash\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.6-sol": {
"id": "openai-gpt-5.6-sol",
"name": "OpenAI GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 4,
"output": 20,
"cache_read": 0.4,
"tiers": [
{
"input": 8,
"output": 30,
"cache_read": 0.8,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 8,
"output": 30,
"cache_read": 0.8
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.6-sol\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-nemo-instruct-2407": {
"id": "mistral-nemo-instruct-2407",
"name": "Mistral Nemo Instruct",
"description": "Legacy model retained for compatibility with older integrations",
"family": "mistral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"status": "deprecated",
"cost": {
"input": 0.3,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/mistral-nemo-instruct-2407\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"mistral-nemo-instruct-2407\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-ultra-550b": {
"id": "nemotron-3-ultra-550b",
"name": "Nemotron 3 Ultra",
"description": "Flagship Nemotron model for high-throughput reasoning and complex agents",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.9,
"output": 1.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/nemotron-3-ultra-550b\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"nemotron-3-ultra-550b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-fable-5.1": {
"id": "anthropic-claude-fable-5.1",
"name": "Anthropic Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-fable-5.1\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-fable-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-nano-12b-v2-vl": {
"id": "nemotron-nano-12b-v2-vl",
"name": "Nemotron-nano 12b v2-vl",
"description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
"family": "nemotron",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-10-28",
"last_updated": "2025-10-28",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.2,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/nemotron-nano-12b-v2-vl\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"nemotron-nano-12b-v2-vl\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM5.3 Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/glm-5.3-flash\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1-distill-llama-70b": {
"id": "deepseek-r1-distill-llama-70b",
"name": "DeepSeek R1 Distill Llama 70B",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-01-30",
"last_updated": "2025-01-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32678,
"output": 8192
},
"cost": {
"input": 0.99,
"output": 0.99
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/deepseek-r1-distill-llama-70b\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-r1-distill-llama-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-o3-mini": {
"id": "openai-o3-mini",
"name": "OpenAI o3 mini",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2024-12-20",
"last_updated": "2025-01-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-o3-mini\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.1-codex-max": {
"id": "openai-gpt-5.1-codex-max",
"name": "GPT-5.1 Codex Max",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.1-codex-max\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.1-codex-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-nano-omni": {
"id": "nemotron-3-nano-omni",
"name": "Nemotron 3 Nano Omni",
"description": "Open Nemotron omni model combining reasoning with text, vision, and audio",
"family": "nemotron",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-28",
"last_updated": "2026-04-28",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/nemotron-3-nano-omni\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"nemotron-3-nano-omni\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.4": {
"id": "openai-gpt-5.4",
"name": "OpenAI GPT-5.4",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.4\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5-nano": {
"id": "openai-gpt-5-nano",
"name": "OpenAI GPT-5 Nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5-nano\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-4.5-sonnet": {
"id": "anthropic-claude-4.5-sonnet",
"name": "Anthropic Claude 4.5 Sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75,
"tiers": [
{
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-4.5-sonnet\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-4.5-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-7b-instruct-v0.3": {
"id": "mistral-7b-instruct-v0.3",
"name": "Mistral 7B Instruct v0.3",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-05-22",
"last_updated": "2024-05-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/mistral-7b-instruct-v0.3\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"mistral-7b-instruct-v0.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ministral-3-8b-instruct-2512": {
"id": "ministral-3-8b-instruct-2512",
"name": "Ministral 3 8B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-12-15",
"last_updated": "2025-12-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/ministral-3-8b-instruct-2512\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"ministral-3-8b-instruct-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.5": {
"id": "openai-gpt-5.5",
"name": "OpenAI GPT-5.5",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.5\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5": {
"id": "glm-5",
"name": "GLM 5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-11",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 64000,
"output": 64000
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/glm-5\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5-mini": {
"id": "openai-gpt-5-mini",
"name": "OpenAI GPT-5 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5-mini\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-3.5-haiku": {
"id": "anthropic-claude-3.5-haiku",
"name": "Claude 3.5 Haiku",
"description": "Legacy model retained for compatibility with older integrations",
"family": "claude-haiku",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-11-05",
"last_updated": "2024-11-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"status": "deprecated",
"cost": {
"input": 0.8,
"output": 4,
"cache_read": 0.08,
"cache_write": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-3.5-haiku\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-3.5-haiku\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max": {
"id": "qwen3.8-max",
"name": "Qwen3.8-Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 262144
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/qwen3.8-max\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "Kimi K2.5",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.5,
"output": 2.7,
"cache_read": 0.203
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/kimi-k2.5\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 1.3,
"output": 4.3,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/glm-5.1\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-3-14B": {
"id": "mistral-3-14B",
"name": "Ministral 3 14B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-15",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/mistral-3-14B\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"mistral-3-14B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-tts-voicedesign": {
"id": "qwen3-tts-voicedesign",
"name": "Qwen3 TTS VoiceDesign",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-04-21",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/qwen3-tts-voicedesign\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"qwen3-tts-voicedesign\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bge-m3": {
"id": "bge-m3",
"name": "BGE M3",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "bge",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-01-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 1024
},
"cost": {
"input": 0.02,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/bge-m3\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"bge-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-embedding-0.6b": {
"id": "qwen3-embedding-0.6b",
"name": "Qwen3 Embedding 0.6B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-06-03",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8000,
"output": 1024
},
"status": "beta",
"cost": {
"input": 0.04,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/qwen3-embedding-0.6b\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"qwen3-embedding-0.6b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-o1": {
"id": "openai-o1",
"name": "OpenAI o1",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2023-09",
"release_date": "2024-12-05",
"last_updated": "2024-12-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 15,
"output": 60,
"cache_read": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-o1\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-o1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"wan2-2-t2v-a14b": {
"id": "wan2-2-t2v-a14b",
"name": "Wan2.2-T2V-A14B",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-07-28",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": true,
"limit": {
"context": 100,
"output": 1
},
"cost": {
"input": 0.6,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/wan2-2-t2v-a14b\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"wan2-2-t2v-a14b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-3-opus": {
"id": "anthropic-claude-3-opus",
"name": "Claude 3 Opus",
"description": "Legacy model retained for compatibility with older integrations",
"family": "claude-opus",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-08",
"release_date": "2024-02-29",
"last_updated": "2024-02-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 4096
},
"status": "deprecated",
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-3-opus\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-3-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-4.1-opus": {
"id": "anthropic-claude-4.1-opus",
"name": "Anthropic Claude 4.1 Opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-4.1-opus\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-4.1-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-image-1": {
"id": "openai-gpt-image-1",
"name": "GPT Image 1",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-image",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-04-24",
"last_updated": "2025-04-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 5,
"output": 40,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-image-1\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-image-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "Deepseek V4 Pro",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.87,
"output": 1.74,
"cache_read": 0.174
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/deepseek-v4-pro\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stable-diffusion-3.5-large": {
"id": "stable-diffusion-3.5-large",
"name": "Stable Diffusion 3.5 Large",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "stable-diffusion",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-10-22",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": true,
"limit": {
"context": 256,
"output": 1
},
"cost": {
"input": 0.08,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/stable-diffusion-3.5-large\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"stable-diffusion-3.5-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-opus-4.5": {
"id": "anthropic-claude-opus-4.5",
"name": "Anthropic Claude Opus 4.5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-opus-4.5\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-opus-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama3-8b-instruct": {
"id": "llama3-8b-instruct",
"name": "Llama 3.1 Instruct (8B)",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.198,
"output": 0.198
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/llama3-8b-instruct\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"llama3-8b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.95,
"output": 3.4,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/glm-5.3\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-opus-5": {
"id": "anthropic-claude-opus-5",
"name": "Anthropic Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-opus-5\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-5.4-nano": {
"id": "openai-gpt-5.4-nano",
"name": "OpenAI GPT-5.4 Nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-5.4-nano\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-4.6-sonnet": {
"id": "anthropic-claude-4.6-sonnet",
"name": "Anthropic Claude Sonnet 4.6",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75,
"tiers": [
{
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-4.6-sonnet\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-4.6-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic-claude-haiku-4.5": {
"id": "anthropic-claude-haiku-4.5",
"name": "Anthropic Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/anthropic-claude-haiku-4.5\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"anthropic-claude-haiku-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-pro": {
"id": "mimo-v2.5-pro",
"name": "MiMo V2.5 Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.4,
"output": 1.5,
"cache_read": 0.08
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/mimo-v2.5-pro\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-image-2": {
"id": "openai-gpt-image-2",
"name": "OpenAI GPT Image 2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-image",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-04-24",
"last_updated": "2025-04-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image",
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 16384
},
"cost": {
"input": 8,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-image-2\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-image-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-4o-mini": {
"id": "openai-gpt-4o-mini",
"name": "OpenAI GPT-4o mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/openai-gpt-4o-mini\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"openai-gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fal-ai/fast-sdxl": {
"id": "fal-ai/fast-sdxl",
"name": "Fast SDXL",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "stable-diffusion",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2023-07-26",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/fal-ai/fast-sdxl\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"fal-ai/fast-sdxl\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fal-ai/elevenlabs/tts/multilingual-v2": {
"id": "fal-ai/elevenlabs/tts/multilingual-v2",
"name": "ElevenLabs Multilingual TTS v2",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "elevenlabs",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2023-08-22",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/fal-ai/elevenlabs/tts/multilingual-v2\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"fal-ai/elevenlabs/tts/multilingual-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fal-ai/stable-audio-25/text-to-audio": {
"id": "fal-ai/stable-audio-25/text-to-audio",
"name": "Stable Audio 2.5 (Text-to-Audio)",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-10-08",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/fal-ai/stable-audio-25/text-to-audio\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"fal-ai/stable-audio-25/text-to-audio\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fal-ai/flux/schnell": {
"id": "fal-ai/flux/schnell",
"name": "FLUX.1 [schnell]",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-08-01",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"digitalocean/fal-ai/flux/schnell\", apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.do-ai.run/v1\")!,\n apiKey: processEnvironment[\"DIGITALOCEAN_ACCESS_TOKEN\"]\n)\nlet session = provider.model(\"fal-ai/flux/schnell\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"vivgrid": {
"id": "vivgrid",
"name": "Vivgrid",
"baseURL": "https://api.vivgrid.com/v1",
"npm": "@ai-sdk/openai",
"swiftDriver": "openaiChat",
"env": [
"VIVGRID_API_KEY"
],
"doc": "https://docs.vivgrid.com/models",
"modelCount": 27,
"models": {
"deepseek-v4-pro-0813": {
"id": "deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 1.35,
"output": 3,
"reasoning": 3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex": {
"id": "gpt-5.1-codex",
"name": "GPT-5.1 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5.1-codex\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-sol": {
"id": "gpt-5.6-sol",
"name": "GPT 5.6 Sol",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5.6-sol\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro-preview": {
"id": "gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2-codex": {
"id": "gpt-5.2-codex",
"name": "GPT-5.2 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-01-14",
"last_updated": "2026-01-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5.2-codex\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.2-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.2,
"output": 4.2,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/glm-5.2\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-6-astra": {
"id": "gpt-6-astra",
"name": "GPT-6 Astra",
"description": "GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.",
"family": "gpt-astra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-04-30",
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5,
"tiers": [
{
"input": 20,
"output": 75,
"cache_read": 2,
"cache_write": 25,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 20,
"output": 75,
"cache_read": 2,
"cache_write": 25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-6-astra\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-6-astra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.15,
"output": 0.3,
"reasoning": 0.3,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/deepseek-v4-flash\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5.4\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5-1": {
"id": "claude-fable-5-1",
"name": "Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.5,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/claude-fable-5-1\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex-max": {
"id": "gpt-5.1-codex-max",
"name": "GPT-5.1 Codex Max",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5.1-codex-max\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-codex-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite-preview": {
"id": "gemini-3.1-flash-lite-preview",
"name": "Gemini 3.1 Flash Lite Preview",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"cache_write": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gemini-3.1-flash-lite-preview\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-luna": {
"id": "gpt-5.6-luna",
"name": "GPT 5.6 Luna",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 1,
"output": 6,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5.6-luna\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/kimi-k3\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3.2": {
"id": "deepseek-v3.2",
"name": "DeepSeek-V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.28,
"output": 0.42
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/deepseek-v3.2\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.3-codex": {
"id": "gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-02-24",
"last_updated": "2026-02-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5.3-codex\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/glm-5.3-flash\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/claude-fable-5\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-nano": {
"id": "gpt-5.4-nano",
"name": "GPT-5.4 Nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5.4-nano\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-mini": {
"id": "gpt-5.4-mini",
"name": "GPT-5.4 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5.4-mini\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.8-flash": {
"id": "gemini-3.8-flash",
"name": "Gemini 3.8 Flash",
"description": "Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gemini-3.8-flash\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.003625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/deepseek-v4-pro\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-mini": {
"id": "gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5-mini\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.7-flash": {
"id": "gemini-3.7-flash",
"name": "Gemini 3.7 Flash",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gemini-3.7-flash\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-terra": {
"id": "gpt-5.6-terra",
"name": "GPT 5.6 Terra",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5.6-terra\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4.2,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/glm-5.3\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vivgrid/gpt-5.5\", apiKey: processEnvironment[\"VIVGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vivgrid.com/v1\")!,\n apiKey: processEnvironment[\"VIVGRID_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"auriko": {
"id": "auriko",
"name": "Auriko",
"baseURL": "https://api.auriko.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"AURIKO_API_KEY"
],
"doc": "https://docs.auriko.ai",
"modelCount": 15,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/claude-sonnet-4-6\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2-7-highspeed": {
"id": "minimax-m2-7-highspeed",
"name": "MiniMax-M2.7-highspeed",
"description": "Low-latency M2.7 variant for interactive coding plans and agent loops",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/minimax-m2-7-highspeed\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2-7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.3": {
"id": "grok-4.3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/grok-4.3\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/kimi-k2.6\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro-preview": {
"id": "gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-3.6-plus": {
"id": "qwen-3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.1,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/qwen-3.6-plus\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"qwen-3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/deepseek-v4-flash\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/claude-opus-4-6\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7": {
"id": "claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 30,
"output": 150,
"cache_read": 3,
"cache_write": 37.5
},
"provider": {
"body": {
"speed": "fast"
},
"headers": {
"anthropic-beta": "fast-mode-2026-02-01"
}
}
}
}
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/claude-opus-4-7\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.5,
"output": 2.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/kimi-k2.5\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/glm-5.1\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.003625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/deepseek-v4-pro\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/gemini-2.5-pro\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2-7": {
"id": "minimax-m2-7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/minimax-m2-7\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"auriko/gemini-2.5-flash\", apiKey: processEnvironment[\"AURIKO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.auriko.ai/v1\")!,\n apiKey: processEnvironment[\"AURIKO_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"siliconflow-cn": {
"id": "siliconflow-cn",
"name": "SiliconFlow (China)",
"baseURL": "https://api.siliconflow.cn/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"SILICONFLOW_CN_API_KEY"
],
"doc": "https://cloud.siliconflow.com/models",
"modelCount": 47,
"models": {
"baidu/ERNIE-4.5-300B-A47B": {
"id": "baidu/ERNIE-4.5-300B-A47B",
"name": "baidu/ERNIE-4.5-300B-A47B",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"family": "ernie",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-02",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131000,
"output": 131000
},
"cost": {
"input": 0.28,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/baidu/ERNIE-4.5-300B-A47B\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"baidu/ERNIE-4.5-300B-A47B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun-ai/Step-3.5-Flash": {
"id": "stepfun-ai/Step-3.5-Flash",
"name": "stepfun-ai/Step-3.5-Flash",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"family": "step",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-11",
"last_updated": "2026-02-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.1,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/stepfun-ai/Step-3.5-Flash\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"stepfun-ai/Step-3.5-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Flash": {
"id": "deepseek-ai/DeepSeek-V4-Flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/deepseek-ai/DeepSeek-V4-Flash\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.1-Terminus": {
"id": "deepseek-ai/DeepSeek-V3.1-Terminus",
"name": "deepseek-ai/DeepSeek-V3.1-Terminus",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-29",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 164000,
"output": 164000
},
"cost": {
"input": 0.27,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/deepseek-ai/DeepSeek-V3.1-Terminus\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.1-Terminus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-OCR": {
"id": "deepseek-ai/DeepSeek-OCR",
"name": "deepseek-ai/DeepSeek-OCR",
"description": "OCR model for extracting structured text from documents and screenshots",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-10-20",
"last_updated": "2025-10-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/deepseek-ai/DeepSeek-OCR\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-OCR\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-R1": {
"id": "deepseek-ai/DeepSeek-R1",
"name": "deepseek-ai/DeepSeek-R1",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-05-28",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 164000,
"output": 164000
},
"cost": {
"input": 0.5,
"output": 2.18
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/deepseek-ai/DeepSeek-R1\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-R1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.2": {
"id": "deepseek-ai/DeepSeek-V3.2",
"name": "deepseek-ai/DeepSeek-V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-03",
"last_updated": "2025-12-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 164000,
"output": 164000
},
"cost": {
"input": 0.27,
"output": 0.42
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/deepseek-ai/DeepSeek-V3.2\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro": {
"id": "deepseek-ai/DeepSeek-V4-Pro",
"name": "deepseek-ai/DeepSeek-V4-Pro",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1049000,
"output": 393000
},
"cost": {
"input": 1.74,
"output": 3.48,
"cache_read": 0.145
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/deepseek-ai/DeepSeek-V4-Pro\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3": {
"id": "deepseek-ai/DeepSeek-V3",
"name": "deepseek-ai/DeepSeek-V3",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-12-26",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 164000,
"output": 164000
},
"cost": {
"input": 0.25,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/deepseek-ai/DeepSeek-V3\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionAI/Ling-flash-2.0": {
"id": "inclusionAI/Ling-flash-2.0",
"name": "inclusionAI/Ling-flash-2.0",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-18",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131000,
"output": 131000
},
"cost": {
"input": 0.14,
"output": 0.57
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/inclusionAI/Ling-flash-2.0\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"inclusionAI/Ling-flash-2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2": {
"id": "zai-org/GLM-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1049000,
"output": 262000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/zai-org/GLM-5.2\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.5-Air": {
"id": "zai-org/GLM-4.5-Air",
"name": "zai-org/GLM-4.5-Air",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm-air",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-28",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131000,
"output": 131000
},
"cost": {
"input": 0.14,
"output": 0.86
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/zai-org/GLM-4.5-Air\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.5-Air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-27B": {
"id": "Qwen/Qwen3.5-27B",
"name": "Qwen/Qwen3.5-27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-25",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.26,
"output": 2.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3.5-27B\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-8B": {
"id": "Qwen/Qwen3-8B",
"name": "Qwen/Qwen3-8B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-30",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131000,
"output": 131000
},
"cost": {
"input": 0.06,
"output": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-8B\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-8B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-14B": {
"id": "Qwen/Qwen3-14B",
"name": "Qwen/Qwen3-14B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-30",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131000,
"output": 131000
},
"cost": {
"input": 0.07,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-14B\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-14B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-4B": {
"id": "Qwen/Qwen3.5-4B",
"name": "Qwen/Qwen3.5-4B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3.5-4B\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-4B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-9B": {
"id": "Qwen/Qwen3.5-9B",
"name": "Qwen/Qwen3.5-9B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.22,
"output": 1.74
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3.5-9B\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-9B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-122B-A10B": {
"id": "Qwen/Qwen3.5-122B-A10B",
"name": "Qwen/Qwen3.5-122B-A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-26",
"last_updated": "2026-02-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.29,
"output": 2.32
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3.5-122B-A10B\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-122B-A10B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-397B-A17B": {
"id": "Qwen/Qwen3.5-397B-A17B",
"name": "Qwen/Qwen3.5-397B-A17B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.29,
"output": 1.74
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3.5-397B-A17B\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-397B-A17B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-35B-A3B": {
"id": "Qwen/Qwen3.5-35B-A3B",
"name": "Qwen/Qwen3.5-35B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-25",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.23,
"output": 1.86
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3.5-35B-A3B\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-32B": {
"id": "Qwen/Qwen3-32B",
"name": "Qwen/Qwen3-32B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-30",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131000,
"output": 131000
},
"cost": {
"input": 0.14,
"output": 0.57
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-32B\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-32B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B-Thinking-2507": {
"id": "Qwen/Qwen3-235B-A22B-Thinking-2507",
"name": "Qwen/Qwen3-235B-A22B-Thinking-2507",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-28",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.13,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-235B-A22B-Thinking-2507\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B-Thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-35B-A3B": {
"id": "Qwen/Qwen3.6-35B-A3B",
"name": "Qwen/Qwen3.6-35B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.23,
"output": 1.86
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3.6-35B-A3B\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-30B-A3B-Instruct": {
"id": "Qwen/Qwen3-VL-30B-A3B-Instruct",
"name": "Qwen/Qwen3-VL-30B-A3B-Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-05",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.29,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-VL-30B-A3B-Instruct\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-30B-A3B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-235B-A22B-Thinking": {
"id": "Qwen/Qwen3-VL-235B-A22B-Thinking",
"name": "Qwen/Qwen3-VL-235B-A22B-Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-04",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.45,
"output": 3.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-VL-235B-A22B-Thinking\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-235B-A22B-Thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-8B-Instruct": {
"id": "Qwen/Qwen3-VL-8B-Instruct",
"name": "Qwen/Qwen3-VL-8B-Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-15",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.18,
"output": 0.68
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-VL-8B-Instruct\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-8B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-32B-Instruct": {
"id": "Qwen/Qwen3-VL-32B-Instruct",
"name": "Qwen/Qwen3-VL-32B-Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-21",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.2,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-VL-32B-Instruct\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-32B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen2.5-72B-Instruct": {
"id": "Qwen/Qwen2.5-72B-Instruct",
"name": "Qwen/Qwen2.5-72B-Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-09-18",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 33000,
"output": 4000
},
"cost": {
"input": 0.59,
"output": 0.59
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen2.5-72B-Instruct\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen2.5-72B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen2.5-7B-Instruct": {
"id": "Qwen/Qwen2.5-7B-Instruct",
"name": "Qwen/Qwen2.5-7B-Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-09-18",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 33000,
"output": 4000
},
"cost": {
"input": 0.05,
"output": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen2.5-7B-Instruct\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen2.5-7B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Coder-480B-A35B-Instruct": {
"id": "Qwen/Qwen3-Coder-480B-A35B-Instruct",
"name": "Qwen/Qwen3-Coder-480B-A35B-Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-31",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.25,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-Coder-480B-A35B-Instruct\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Coder-480B-A35B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-235B-A22B-Instruct": {
"id": "Qwen/Qwen3-VL-235B-A22B-Instruct",
"name": "Qwen/Qwen3-VL-235B-A22B-Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-04",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.3,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-VL-235B-A22B-Instruct\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-235B-A22B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Coder-30B-A3B-Instruct": {
"id": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
"name": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-01",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.07,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-Coder-30B-A3B-Instruct\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Coder-30B-A3B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-32B-Thinking": {
"id": "Qwen/Qwen3-VL-32B-Thinking",
"name": "Qwen/Qwen3-VL-32B-Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-21",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.2,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-VL-32B-Thinking\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-32B-Thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-30B-A3B-Thinking": {
"id": "Qwen/Qwen3-VL-30B-A3B-Thinking",
"name": "Qwen/Qwen3-VL-30B-A3B-Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-11",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.29,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-VL-30B-A3B-Thinking\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-30B-A3B-Thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-30B-A3B-Instruct-2507": {
"id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
"name": "Qwen/Qwen3-30B-A3B-Instruct-2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-30",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.09,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Qwen/Qwen3-30B-A3B-Instruct-2507\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-30B-A3B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ByteDance-Seed/Seed-OSS-36B-Instruct": {
"id": "ByteDance-Seed/Seed-OSS-36B-Instruct",
"name": "ByteDance-Seed/Seed-OSS-36B-Instruct",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-04",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.21,
"output": 0.57
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/ByteDance-Seed/Seed-OSS-36B-Instruct\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"ByteDance-Seed/Seed-OSS-36B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Pro/deepseek-ai/DeepSeek-V3": {
"id": "Pro/deepseek-ai/DeepSeek-V3",
"name": "Pro/deepseek-ai/DeepSeek-V3",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-12-26",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 164000,
"output": 164000
},
"cost": {
"input": 0.25,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Pro/deepseek-ai/DeepSeek-V3\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Pro/deepseek-ai/DeepSeek-V3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Pro/deepseek-ai/DeepSeek-V3.1-Terminus": {
"id": "Pro/deepseek-ai/DeepSeek-V3.1-Terminus",
"name": "Pro/deepseek-ai/DeepSeek-V3.1-Terminus",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-29",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 164000,
"output": 164000
},
"cost": {
"input": 0.27,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Pro/deepseek-ai/DeepSeek-V3.1-Terminus\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Pro/deepseek-ai/DeepSeek-V3.1-Terminus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Pro/deepseek-ai/DeepSeek-R1": {
"id": "Pro/deepseek-ai/DeepSeek-R1",
"name": "Pro/deepseek-ai/DeepSeek-R1",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-05-28",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 164000,
"output": 164000
},
"cost": {
"input": 0.5,
"output": 2.18
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Pro/deepseek-ai/DeepSeek-R1\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Pro/deepseek-ai/DeepSeek-R1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Pro/deepseek-ai/DeepSeek-V3.2": {
"id": "Pro/deepseek-ai/DeepSeek-V3.2",
"name": "Pro/deepseek-ai/DeepSeek-V3.2",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-03",
"last_updated": "2025-12-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 164000,
"output": 164000
},
"cost": {
"input": 0.27,
"output": 0.42
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Pro/deepseek-ai/DeepSeek-V3.2\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Pro/deepseek-ai/DeepSeek-V3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Pro/zai-org/GLM-5.1": {
"id": "Pro/zai-org/GLM-5.1",
"name": "Pro/zai-org/GLM-5.1",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-08",
"last_updated": "2026-04-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 205000,
"output": 205000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Pro/zai-org/GLM-5.1\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Pro/zai-org/GLM-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Pro/zai-org/GLM-5": {
"id": "Pro/zai-org/GLM-5",
"name": "Pro/zai-org/GLM-5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 205000,
"output": 205000
},
"cost": {
"input": 1,
"output": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Pro/zai-org/GLM-5\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Pro/zai-org/GLM-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Pro/MiniMaxAI/MiniMax-M2.5": {
"id": "Pro/MiniMaxAI/MiniMax-M2.5",
"name": "Pro/MiniMaxAI/MiniMax-M2.5",
"description": "Frontier MiniMax model for engineering, office tasks, and agentic reasoning",
"family": "minimax",
"attachment": false,
"reasoning": false,
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-13",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 192000,
"output": 131000
},
"cost": {
"input": 0.3,
"output": 1.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Pro/MiniMaxAI/MiniMax-M2.5\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Pro/MiniMaxAI/MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Pro/moonshotai/Kimi-K2.5": {
"id": "Pro/moonshotai/Kimi-K2.5",
"name": "Pro/moonshotai/Kimi-K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.45,
"output": 2.25,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Pro/moonshotai/Kimi-K2.5\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Pro/moonshotai/Kimi-K2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Pro/moonshotai/Kimi-K2.6": {
"id": "Pro/moonshotai/Kimi-K2.6",
"name": "Pro/moonshotai/Kimi-K2.6",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/Pro/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"Pro/moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/Hunyuan-A13B-Instruct": {
"id": "tencent/Hunyuan-A13B-Instruct",
"name": "tencent/Hunyuan-A13B-Instruct",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "hunyuan",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-06-30",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131000,
"output": 131000
},
"cost": {
"input": 0.14,
"output": 0.57
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/tencent/Hunyuan-A13B-Instruct\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"tencent/Hunyuan-A13B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"PaddlePaddle/PaddleOCR-VL-1.5": {
"id": "PaddlePaddle/PaddleOCR-VL-1.5",
"name": "PaddlePaddle/PaddleOCR-VL-1.5",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-01-29",
"last_updated": "2026-01-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"siliconflow-cn/PaddlePaddle/PaddleOCR-VL-1.5\", apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.siliconflow.cn/v1\")!,\n apiKey: processEnvironment[\"SILICONFLOW_CN_API_KEY\"]\n)\nlet session = provider.model(\"PaddlePaddle/PaddleOCR-VL-1.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"nova": {
"id": "nova",
"name": "Nova",
"baseURL": "https://api.nova.amazon.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"NOVA_API_KEY"
],
"doc": "https://nova.amazon.com/dev/documentation",
"modelCount": 2,
"models": {
"nova-2-lite-v1": {
"id": "nova-2-lite-v1",
"name": "Nova 2 Lite",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "nova-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0,
"output": 0,
"reasoning": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nova/nova-2-lite-v1\", apiKey: processEnvironment[\"NOVA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.nova.amazon.com/v1\")!,\n apiKey: processEnvironment[\"NOVA_API_KEY\"]\n)\nlet session = provider.model(\"nova-2-lite-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nova-2-pro-v1": {
"id": "nova-2-pro-v1",
"name": "Nova 2 Pro",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "nova-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-03",
"last_updated": "2026-01-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0,
"output": 0,
"reasoning": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nova/nova-2-pro-v1\", apiKey: processEnvironment[\"NOVA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.nova.amazon.com/v1\")!,\n apiKey: processEnvironment[\"NOVA_API_KEY\"]\n)\nlet session = provider.model(\"nova-2-pro-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"inceptron": {
"id": "inceptron",
"name": "Inceptron",
"baseURL": "https://api.inceptron.io/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"INCEPTRON_API_KEY"
],
"doc": "https://docs.inceptron.io",
"modelCount": 4,
"models": {
"deepseek-ai/DeepSeek-V4-Flash-0731": {
"id": "deepseek-ai/DeepSeek-V4-Flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.13,
"output": 0.28,
"cache_read": 0.03,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inceptron/deepseek-ai/DeepSeek-V4-Flash-0731\", apiKey: processEnvironment[\"INCEPTRON_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inceptron.io/v1\")!,\n apiKey: processEnvironment[\"INCEPTRON_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2": {
"id": "zai-org/GLM-5.2",
"name": "GLM 5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.71,
"output": 2.35,
"cache_read": 0.12,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inceptron/zai-org/GLM-5.2\", apiKey: processEnvironment[\"INCEPTRON_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inceptron.io/v1\")!,\n apiKey: processEnvironment[\"INCEPTRON_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.7-Code": {
"id": "moonshotai/Kimi-K2.7-Code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.66,
"output": 3.4,
"cache_read": 0.18,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inceptron/moonshotai/Kimi-K2.7-Code\", apiKey: processEnvironment[\"INCEPTRON_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inceptron.io/v1\")!,\n apiKey: processEnvironment[\"INCEPTRON_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.7-Code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6": {
"id": "moonshotai/Kimi-K2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.53,
"output": 3.39,
"cache_read": 0.17,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inceptron/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"INCEPTRON_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inceptron.io/v1\")!,\n apiKey: processEnvironment[\"INCEPTRON_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"vultr": {
"id": "vultr",
"name": "Vultr",
"baseURL": "https://api.vultrinference.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"VULTR_API_KEY"
],
"doc": "https://api.vultrinference.com/",
"modelCount": 10,
"models": {
"deepseek-ai/DeepSeek-V4-Flash": {
"id": "deepseek-ai/DeepSeek-V4-Flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.3,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vultr/deepseek-ai/DeepSeek-V4-Flash\", apiKey: processEnvironment[\"VULTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vultrinference.com/v1\")!,\n apiKey: processEnvironment[\"VULTR_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/DeepSeek-V3.2-NVFP4": {
"id": "nvidia/DeepSeek-V3.2-NVFP4",
"name": "DeepSeek V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.55,
"output": 1.65
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vultr/nvidia/DeepSeek-V3.2-NVFP4\", apiKey: processEnvironment[\"VULTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vultrinference.com/v1\")!,\n apiKey: processEnvironment[\"VULTR_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/DeepSeek-V3.2-NVFP4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16": {
"id": "nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16",
"name": "NVIDIA Nemotron 3 Nano Omni",
"description": "Open Nemotron omni model combining reasoning with text, vision, and audio",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-28",
"last_updated": "2026-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.13,
"output": 0.38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vultr/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16\", apiKey: processEnvironment[\"VULTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vultrinference.com/v1\")!,\n apiKey: processEnvironment[\"VULTR_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Nemotron-Cascade-2-30B-A3B": {
"id": "nvidia/Nemotron-Cascade-2-30B-A3B",
"name": "NVIDIA Nemotron Cascade 2",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vultr/nvidia/Nemotron-Cascade-2-30B-A3B\", apiKey: processEnvironment[\"VULTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vultrinference.com/v1\")!,\n apiKey: processEnvironment[\"VULTR_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Nemotron-Cascade-2-30B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2-FP8": {
"id": "zai-org/GLM-5.2-FP8",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 393216,
"output": 131072
},
"cost": {
"input": 0.85,
"output": 3.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vultr/zai-org/GLM-5.2-FP8\", apiKey: processEnvironment[\"VULTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vultrinference.com/v1\")!,\n apiKey: processEnvironment[\"VULTR_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-397B-A17B": {
"id": "Qwen/Qwen3.5-397B-A17B",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vultr/Qwen/Qwen3.5-397B-A17B\", apiKey: processEnvironment[\"VULTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vultrinference.com/v1\")!,\n apiKey: processEnvironment[\"VULTR_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-397B-A17B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-27B": {
"id": "Qwen/Qwen3.6-27B",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vultr/Qwen/Qwen3.6-27B\", apiKey: processEnvironment[\"VULTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vultrinference.com/v1\")!,\n apiKey: processEnvironment[\"VULTR_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M2.7": {
"id": "MiniMaxAI/MiniMax-M2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vultr/MiniMaxAI/MiniMax-M2.7\", apiKey: processEnvironment[\"VULTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vultrinference.com/v1\")!,\n apiKey: processEnvironment[\"VULTR_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6": {
"id": "moonshotai/Kimi-K2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vultr/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"VULTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vultrinference.com/v1\")!,\n apiKey: processEnvironment[\"VULTR_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"XiaomiMiMo/MiMo-V2.5-Pro": {
"id": "XiaomiMiMo/MiMo-V2.5-Pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.55,
"output": 1.65
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vultr/XiaomiMiMo/MiMo-V2.5-Pro\", apiKey: processEnvironment[\"VULTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.vultrinference.com/v1\")!,\n apiKey: processEnvironment[\"VULTR_API_KEY\"]\n)\nlet session = provider.model(\"XiaomiMiMo/MiMo-V2.5-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"ollama-cloud": {
"id": "ollama-cloud",
"name": "Ollama Cloud",
"baseURL": "https://ollama.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "ollamaNative",
"env": [
"OLLAMA_API_KEY"
],
"doc": "https://docs.ollama.com/cloud",
"modelCount": 23,
"models": {
"gpt-oss:20b": {
"id": "gpt-oss:20b",
"name": "gpt-oss:20b",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"release_date": "2025-08-05",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/gpt-oss:20b\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss:20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash:0731": {
"id": "deepseek-v4-flash:0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/deepseek-v4-flash:0731\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash:0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.7": {
"id": "minimax-m2.7",
"name": "minimax-m2.7",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 196608
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/minimax-m2.7\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "kimi-k2.6",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"release_date": "2026-04-20",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/kimi-k2.6\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 976000,
"output": 131072
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/glm-5.2\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.5": {
"id": "minimax-m2.5",
"name": "minimax-m2.5",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"knowledge": "2025-01",
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/minimax-m2.5\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m3": {
"id": "minimax-m3",
"name": "minimax-m3",
"description": "MiniMax multimodal coding model for long-context reasoning and agent tasks",
"family": "minimax-m3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-31",
"last_updated": "2026-05-31",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512000,
"output": 131072
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/minimax-m3\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5:397b": {
"id": "qwen3.5:397b",
"name": "qwen3.5:397b",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_details"
},
"release_date": "2026-02-15",
"last_updated": "2026-02-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/qwen3.5:397b\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5:397b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "deepseek-v4-flash",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/deepseek-v4-flash\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code": {
"id": "kimi-k2.7-code",
"name": "kimi-k2.7-code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/kimi-k2.7-code\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss:120b": {
"id": "gpt-oss:120b",
"name": "gpt-oss:120b",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"release_date": "2025-08-05",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/gpt-oss:120b\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss:120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-ultra": {
"id": "nemotron-3-ultra",
"name": "nemotron-3-ultra",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 128000
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/nemotron-3-ultra\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4.1-flash": {
"id": "deepseek-v4.1-flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/deepseek-v4.1-flash\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4.1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-nano:30b": {
"id": "nemotron-3-nano:30b",
"name": "nemotron-3-nano:30b",
"description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-15",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/nemotron-3-nano:30b\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-nano:30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-large-3:675b": {
"id": "mistral-large-3:675b",
"name": "mistral-large-3:675b",
"description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
"family": "mistral-large",
"attachment": true,
"reasoning": false,
"tool_call": true,
"release_date": "2025-12-02",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/mistral-large-3:675b\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"mistral-large-3:675b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "kimi-k3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/kimi-k3\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/glm-5.3-flash\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma4:31b": {
"id": "gemma4:31b",
"name": "gemma4:31b",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"knowledge": "2025-01",
"release_date": "2026-04-02",
"last_updated": "2026-04-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/gemma4:31b\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"gemma4:31b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "kimi-k2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/kimi-k2.5\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "glm-5.1",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"release_date": "2026-03-27",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/glm-5.1\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "deepseek-v4-pro",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/deepseek-v4-pro\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/glm-5.3\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-super": {
"id": "nemotron-3-super",
"name": "nemotron-3-super",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"swiftDriver": "ollamaNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ollama-cloud/nemotron-3-super\", apiKey: processEnvironment[\"OLLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ollama.com/v1\")!,\n apiKey: processEnvironment[\"OLLAMA_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-super\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"freemodel": {
"id": "freemodel",
"name": "FreeModel",
"baseURL": "https://cc.freemodel.dev/v1",
"npm": "@ai-sdk/anthropic",
"swiftDriver": "anthropicMessages",
"env": [
"FREEMODEL_API_KEY"
],
"doc": "https://freemodel.dev",
"modelCount": 10,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"freemodel/claude-sonnet-4-6\", apiKey: processEnvironment[\"FREEMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://cc.freemodel.dev/v1\")!,\n apiKey: processEnvironment[\"FREEMODEL_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://api.freemodel.dev/v1"
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"freemodel/gpt-5.4\", apiKey: processEnvironment[\"FREEMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://cc.freemodel.dev/v1\")!,\n apiKey: processEnvironment[\"FREEMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"freemodel/claude-opus-4-6\", apiKey: processEnvironment[\"FREEMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://cc.freemodel.dev/v1\")!,\n apiKey: processEnvironment[\"FREEMODEL_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7": {
"id": "claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"freemodel/claude-opus-4-7\", apiKey: processEnvironment[\"FREEMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://cc.freemodel.dev/v1\")!,\n apiKey: processEnvironment[\"FREEMODEL_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.3-codex": {
"id": "gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://api.freemodel.dev/v1"
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175,
"cache_write": 1.75
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"freemodel/gpt-5.3-codex\", apiKey: processEnvironment[\"FREEMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://cc.freemodel.dev/v1\")!,\n apiKey: processEnvironment[\"FREEMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5-20251001": {
"id": "claude-haiku-4-5-20251001",
"name": "Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"freemodel/claude-haiku-4-5-20251001\", apiKey: processEnvironment[\"FREEMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://cc.freemodel.dev/v1\")!,\n apiKey: processEnvironment[\"FREEMODEL_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5-20251001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"freemodel/claude-fable-5\", apiKey: processEnvironment[\"FREEMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://cc.freemodel.dev/v1\")!,\n apiKey: processEnvironment[\"FREEMODEL_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-mini": {
"id": "gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://api.freemodel.dev/v1"
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075,
"cache_write": 0.75
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"freemodel/gpt-5.4-mini\", apiKey: processEnvironment[\"FREEMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://cc.freemodel.dev/v1\")!,\n apiKey: processEnvironment[\"FREEMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"freemodel/claude-opus-4-8\", apiKey: processEnvironment[\"FREEMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://cc.freemodel.dev/v1\")!,\n apiKey: processEnvironment[\"FREEMODEL_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://api.freemodel.dev/v1"
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 5
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"freemodel/gpt-5.5\", apiKey: processEnvironment[\"FREEMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://cc.freemodel.dev/v1\")!,\n apiKey: processEnvironment[\"FREEMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"iflowcn": {
"id": "iflowcn",
"name": "iFlow",
"baseURL": "https://apis.iflow.cn/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"IFLOW_API_KEY"
],
"doc": "https://platform.iflow.cn/en/docs",
"modelCount": 14,
"models": {
"qwen3-coder-plus": {
"id": "qwen3-coder-plus",
"name": "Qwen3-Coder-Plus",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-01",
"last_updated": "2025-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/qwen3-coder-plus\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3": {
"id": "deepseek-v3",
"name": "DeepSeek-V3",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-12-26",
"last_updated": "2024-12-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/deepseek-v3\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-thinking-2507": {
"id": "qwen3-235b-a22b-thinking-2507",
"name": "Qwen3-235B-A22B-Thinking",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-01",
"last_updated": "2025-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/qwen3-235b-a22b-thinking-2507\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.6": {
"id": "glm-4.6",
"name": "GLM-4.6",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-12-01",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/glm-4.6\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-instruct": {
"id": "qwen3-235b-a22b-instruct",
"name": "Qwen3-235B-A22B-Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-01",
"last_updated": "2025-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/qwen3-235b-a22b-instruct\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b": {
"id": "qwen3-235b",
"name": "Qwen3-235B-A22B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-12-01",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/qwen3-235b\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-0905": {
"id": "kimi-k2-0905",
"name": "Kimi-K2-0905",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-09-05",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/kimi-k2-0905\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-0905\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-32b": {
"id": "qwen3-32b",
"name": "Qwen3-32B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-12-01",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/qwen3-32b\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-plus": {
"id": "qwen3-vl-plus",
"name": "Qwen3-VL-Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/qwen3-vl-plus\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-max-preview": {
"id": "qwen3-max-preview",
"name": "Qwen3-Max-Preview",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/qwen3-max-preview\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1": {
"id": "deepseek-r1",
"name": "DeepSeek-R1",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-01-20",
"last_updated": "2025-01-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/deepseek-r1\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3.2": {
"id": "deepseek-v3.2",
"name": "DeepSeek-V3.2-Exp",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 64000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/deepseek-v3.2\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-max": {
"id": "qwen3-max",
"name": "Qwen3-Max",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/qwen3-max\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2": {
"id": "kimi-k2",
"name": "Kimi-K2",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-12-01",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 64000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"iflowcn/kimi-k2\", apiKey: processEnvironment[\"IFLOW_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://apis.iflow.cn/v1\")!,\n apiKey: processEnvironment[\"IFLOW_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"scx-ai": {
"id": "scx-ai",
"name": "SCX.ai",
"baseURL": "https://api.scx.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"SCX_API_KEY"
],
"doc": "https://platform.scx.ai/docs",
"modelCount": 4,
"models": {
"Qwen3.8-Max": {
"id": "Qwen3.8-Max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 983616,
"output": 131072
},
"cost": {
"input": 1.815,
"output": 5.4461,
"cache_read": 0.17,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scx-ai/Qwen3.8-Max\", apiKey: processEnvironment[\"SCX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scx.ai/v1\")!,\n apiKey: processEnvironment[\"SCX_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3.8-Max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"GLM-5.2": {
"id": "GLM-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.55,
"output": 1.784,
"cache_read": 0.111
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scx-ai/GLM-5.2\", apiKey: processEnvironment[\"SCX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scx.ai/v1\")!,\n apiKey: processEnvironment[\"SCX_API_KEY\"]\n)\nlet session = provider.model(\"GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.17,
"output": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scx-ai/gpt-oss-120b\", apiKey: processEnvironment[\"SCX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scx.ai/v1\")!,\n apiKey: processEnvironment[\"SCX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.7": {
"id": "MiniMax-M2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 196608
},
"cost": {
"input": 0.48,
"output": 1.79,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scx-ai/MiniMax-M2.7\", apiKey: processEnvironment[\"SCX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scx.ai/v1\")!,\n apiKey: processEnvironment[\"SCX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"evroc": {
"id": "evroc",
"name": "evroc",
"baseURL": "https://models.think.evroc.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"EVROC_API_KEY"
],
"doc": "https://docs.evroc.com/products/think/overview.html",
"modelCount": 16,
"models": {
"evroc/roc": {
"id": "evroc/roc",
"name": "roc",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2026-01",
"release_date": "2026-06-06",
"last_updated": "2026-06-06",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 2.875,
"output": 11.516
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/evroc/roc\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"evroc/roc\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Voxtral-Small-24B-2507": {
"id": "mistralai/Voxtral-Small-24B-2507",
"name": "Voxtral Small 24B",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "voxtral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2025-03-01",
"last_updated": "2025-03-01",
"modalities": {
"input": [
"audio",
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 32000
},
"cost": {
"input": 0.0023,
"output": 0.0023,
"output_audio": 2.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/mistralai/Voxtral-Small-24B-2507\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Voxtral-Small-24B-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Mistral-Medium-3.5-128B": {
"id": "mistralai/Mistral-Medium-3.5-128B",
"name": "Mistral Medium 3.5",
"description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
"family": "mistral-medium",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-29",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.725,
"output": 6.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/mistralai/Mistral-Medium-3.5-128B\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Mistral-Medium-3.5-128B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Llama-3.3-70B-Instruct-FP8": {
"id": "nvidia/Llama-3.3-70B-Instruct-FP8",
"name": "Llama-3.3-70B-Instruct",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 1.15,
"output": 1.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/nvidia/Llama-3.3-70B-Instruct-FP8\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Llama-3.3-70B-Instruct-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-26B-A4B-it": {
"id": "google/gemma-4-26B-A4B-it",
"name": "Gemma 4 26B A4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.144,
"output": 0.575
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/google/gemma-4-26B-A4B-it\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-26B-A4B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2": {
"id": "zai-org/GLM-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 131072
},
"cost": {
"input": 1.4375,
"output": 5.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/zai-org/GLM-5.2\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.8-27B": {
"id": "Qwen/Qwen3.8-27B",
"name": "Qwen3.8-27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.87,
"output": 3.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/Qwen/Qwen3.8-27B\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.8-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Reranker-4B": {
"id": "Qwen/Qwen3-Reranker-4B",
"name": "Qwen3 Reranker 4B",
"description": "Reranking model for improving retrieval quality in search and recommendation systems",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2025-07-30",
"last_updated": "2025-07-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 4096
},
"cost": {
"input": 0.0575,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/Qwen/Qwen3-Reranker-4B\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Reranker-4B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-35B-A3B": {
"id": "Qwen/Qwen3.6-35B-A3B",
"name": "Qwen3.6 35B-A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.345,
"output": 1.38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/Qwen/Qwen3.6-35B-A3B\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Embedding-8B": {
"id": "Qwen/Qwen3-Embedding-8B",
"name": "Qwen3 Embedding 8B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2025-07-30",
"last_updated": "2025-07-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 4096
},
"cost": {
"input": 0.115,
"output": 0.115
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/Qwen/Qwen3-Embedding-8B\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Embedding-8B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"intfloat/multilingual-e5-large-instruct": {
"id": "intfloat/multilingual-e5-large-instruct",
"name": "E5 Multi-Lingual Large Embeddings 0.6B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2024-06-01",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512,
"output": 512
},
"cost": {
"input": 0.114,
"output": 0.114
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/intfloat/multilingual-e5-large-instruct\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"intfloat/multilingual-e5-large-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"KBLab/kb-whisper-large": {
"id": "KBLab/kb-whisper-large",
"name": "KB Whisper",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "whisper",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2024-10-01",
"last_updated": "2024-10-01",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 448,
"output": 448
},
"cost": {
"input": 0.0023,
"output": 0.0023,
"output_audio": 2.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/KBLab/kb-whisper-large\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"KBLab/kb-whisper-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/whisper-large-v3": {
"id": "openai/whisper-large-v3",
"name": "Whisper 3 Large",
"description": "Open Whisper checkpoint for robust multilingual transcription and captioning",
"family": "whisper",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2024-10-01",
"last_updated": "2024-10-01",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 448,
"output": 4096
},
"cost": {
"input": 0.0023,
"output": 0.0023,
"output_audio": 2.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/openai/whisper-large-v3\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"openai/whisper-large-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/whisper-large-v3-turbo": {
"id": "openai/whisper-large-v3-turbo",
"name": "Whisper Large v3 Turbo",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "whisper",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2024-10-01",
"last_updated": "2024-10-01",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 448,
"output": 448
},
"cost": {
"input": 0.0023,
"output": 0.0023,
"output_audio": 2.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/openai/whisper-large-v3-turbo\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"openai/whisper-large-v3-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.23,
"output": 0.92
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/openai/gpt-oss-120b\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6": {
"id": "moonshotai/Kimi-K2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.4375,
"output": 5.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"evroc/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"EVROC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://models.think.evroc.com/v1\")!,\n apiKey: processEnvironment[\"EVROC_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"echo": {
"id": "echo",
"name": "Echo",
"baseURL": "https://echo.tracerml.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ECHO_API_KEY"
],
"doc": "https://echo.tracerml.ai/docs/api",
"modelCount": 1,
"models": {
"echo": {
"id": "echo",
"name": "Echo",
"description": "Adaptive model for coding, reasoning, and tool-driven agent workflows through one OpenAI-compatible endpoint",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-19",
"last_updated": "2026-08-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"status": "beta",
"cost": {
"input": 10,
"output": 50
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"echo/echo\", apiKey: processEnvironment[\"ECHO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://echo.tracerml.ai/v1\")!,\n apiKey: processEnvironment[\"ECHO_API_KEY\"]\n)\nlet session = provider.model(\"echo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"aixy": {
"id": "aixy",
"name": "Aixy",
"baseURL": "https://api.aixy-gateway.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"AIXY_API_KEY"
],
"doc": "https://docs.aixy-gateway.com/integrations/overview",
"modelCount": 1,
"models": {
"openai/gpt-4.1-mini": {
"id": "openai/gpt-4.1-mini",
"name": "GPT-4.1 mini",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aixy/openai/gpt-4.1-mini\", apiKey: processEnvironment[\"AIXY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.aixy-gateway.com/v1\")!,\n apiKey: processEnvironment[\"AIXY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"impossibl": {
"id": "impossibl",
"name": "Impossibl",
"baseURL": "https://api.impossibl.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"IMPOSSIBL_API_KEY"
],
"doc": "https://impossibl.com/docs/models",
"modelCount": 76,
"models": {
"qwen/qwen3.7-max": {
"id": "qwen/qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/qwen/qwen3.7-max\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-max-preview": {
"id": "qwen/qwen3.8-max-preview",
"name": "Qwen3.8 Max Preview",
"description": "Preview Qwen flagship for million-token multimodal reasoning and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-19",
"last_updated": "2026-07-19",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2.5,
"output": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/qwen/qwen3.8-max-preview\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-flash": {
"id": "qwen/qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.05,
"tiers": [
{
"input": 1,
"output": 4,
"cache_read": 0.2,
"tier": {
"type": "context",
"size": 262144
}
}
],
"context_over_200k": {
"input": 1,
"output": 4,
"cache_read": 0.2
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/qwen/qwen3.6-flash\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.7-plus": {
"id": "qwen/qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.08,
"tiers": [
{
"input": 1.2,
"output": 4.8,
"cache_read": 0.24,
"tier": {
"type": "context",
"size": 262144
}
}
],
"context_over_200k": {
"input": 1.2,
"output": 4.8,
"cache_read": 0.24
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/qwen/qwen3.7-plus\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"groq/gpt-oss-20b": {
"id": "groq/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.075,
"output": 0.3,
"cache_read": 0.0375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/groq/gpt-oss-20b\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"groq/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"groq/gpt-oss-120b": {
"id": "groq/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/groq/gpt-oss-120b\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"groq/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5": {
"id": "xiaomi/mimo-v2.5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/xiaomi/mimo-v2.5\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4-6": {
"id": "anthropic/claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/anthropic/claude-sonnet-4-6\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-5": {
"id": "anthropic/claude-opus-4-5",
"name": "Claude Opus 4.5 (latest)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/anthropic/claude-opus-4-5\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-6": {
"id": "anthropic/claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/anthropic/claude-opus-4-6\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-7": {
"id": "anthropic/claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/anthropic/claude-opus-4-7\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5": {
"id": "anthropic/claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/anthropic/claude-fable-5\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-4-5": {
"id": "anthropic/claude-haiku-4-5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/anthropic/claude-haiku-4-5\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4-5": {
"id": "anthropic/claude-sonnet-4-5",
"name": "Claude Sonnet 4.5 (latest)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/anthropic/claude-sonnet-4-5\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-8": {
"id": "anthropic/claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/anthropic/claude-opus-4-8\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-5": {
"id": "anthropic/claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/anthropic/claude-sonnet-5\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-pro-preview": {
"id": "google/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/google/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash-lite": {
"id": "google/gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash-Lite",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 512,
"max": 24576
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/google/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.6-flash": {
"id": "google/gemini-3.6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 7.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/google/gemini-3.6-flash\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-flash-lite": {
"id": "google/gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/google/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-flash": {
"id": "google/gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/google/gemini-3.5-flash\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-flash-lite": {
"id": "google/gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/google/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-pro": {
"id": "google/gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/google/gemini-2.5-pro\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash": {
"id": "google/gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/google/gemini-2.5-flash\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/inkling": {
"id": "thinkingmachines/inkling",
"name": "Inkling",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 1.87,
"output": 4.68,
"cache_read": 0.374
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/thinkingmachines/inkling\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash": {
"id": "deepseek/deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.19,
"output": 0.51,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/deepseek/deepseek-v4-flash\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro": {
"id": "deepseek/deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 1.74,
"output": 3.48,
"cache_read": 0.145
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/deepseek/deepseek-v4-pro\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-nano": {
"id": "openai/gpt-5-nano",
"name": "GPT-5 Nano",
"description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5-nano\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-nano": {
"id": "openai/gpt-4.1-nano",
"name": "GPT-4.1 nano",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-4.1-nano\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-codex": {
"id": "openai/gpt-5-codex",
"name": "GPT-5-Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-09-15",
"last_updated": "2025-09-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5-codex\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-pro": {
"id": "openai/gpt-5-pro",
"name": "GPT-5 Pro",
"description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 272000
},
"cost": {
"input": 15,
"output": 120
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5-pro\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex-mini": {
"id": "openai/gpt-5.1-codex-mini",
"name": "GPT-5.1 Codex mini",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.1-codex-mini\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex": {
"id": "openai/gpt-5.1-codex",
"name": "GPT-5.1 Codex",
"description": "Codex GPT for repository edits, code review, and practical software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.1-codex\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-sol": {
"id": "openai/gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.6-sol\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-codex": {
"id": "openai/gpt-5.2-codex",
"name": "GPT-5.2 Codex",
"description": "Code-specialist GPT for repository edits, reviews, and long-running software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.2-codex\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-mini": {
"id": "openai/gpt-4.1-mini",
"name": "GPT-4.1 mini",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-4.1-mini\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4": {
"id": "openai/gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.4\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4-turbo": {
"id": "openai/gpt-4-turbo",
"name": "GPT-4 Turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 10,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-4-turbo\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1": {
"id": "openai/gpt-5.1",
"name": "GPT-5.1",
"description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.1\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o1": {
"id": "openai/o1",
"name": "o1",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2023-09",
"release_date": "2024-12-05",
"last_updated": "2024-12-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 15,
"output": 60,
"cache_read": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/o1\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/o1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o": {
"id": "openai/gpt-4o",
"name": "GPT-4o",
"description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-4o\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-luna": {
"id": "openai/gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25,
"tiers": [
{
"input": 0.4,
"output": 1.8,
"cache_read": 0.04,
"cache_write": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.04,
"cache_write": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.6-luna\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-codex": {
"id": "openai/gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.3-codex\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini": {
"id": "openai/gpt-4o-mini",
"name": "GPT-4o mini",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-4o-mini\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1": {
"id": "openai/gpt-4.1",
"name": "GPT-4.1",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-4.1\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-nano": {
"id": "openai/gpt-5.4-nano",
"name": "GPT-5.4 nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.4-nano\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5-pro": {
"id": "openai/gpt-5.5-pro",
"name": "GPT-5.5 Pro",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180,
"cache_read": 3,
"tiers": [
{
"input": 60,
"output": 270,
"cache_read": 3,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 60,
"output": 270,
"cache_read": 3
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.5-pro\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-mini": {
"id": "openai/gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.4-mini\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-3.5-turbo": {
"id": "openai/gpt-3.5-turbo",
"name": "GPT-3.5-turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2021-09-01",
"release_date": "2023-03-01",
"last_updated": "2023-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16385,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-3.5-turbo\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-3.5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-mini": {
"id": "openai/gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5-mini\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-pro": {
"id": "openai/gpt-5.4-pro",
"name": "GPT-5.4 Pro",
"description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180,
"cache_read": 3,
"tiers": [
{
"input": 60,
"output": 270,
"cache_read": 3,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 60,
"output": 270,
"cache_read": 3
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.4-pro\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-terra": {
"id": "openai/gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.6-terra\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2": {
"id": "openai/gpt-5.2",
"name": "GPT-5.2",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.2\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5": {
"id": "openai/gpt-5",
"name": "GPT-5",
"description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o4-mini": {
"id": "openai/o4-mini",
"name": "o4-mini",
"description": "Fast o-series model for compact reasoning, coding, and tool use",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/o4-mini\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/o4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-mini": {
"id": "openai/o3-mini",
"name": "o3-mini",
"description": "Smaller o-series reasoner for economical coding, math, and planning tasks",
"family": "o-mini",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2024-12-20",
"last_updated": "2025-01-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/o3-mini\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3": {
"id": "openai/o3",
"name": "o3",
"description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/o3\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/openai/gpt-5.5\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k3": {
"id": "moonshotai/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/moonshotai/kimi-k3\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.3": {
"id": "xai/grok-4.3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/xai/grok-4.3\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.20-0309-reasoning": {
"id": "xai/grok-4.20-0309-reasoning",
"name": "Grok 4.20 (Reasoning)",
"description": "Reasoning Grok for document-heavy analysis and long-horizon tool use",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/xai/grok-4.20-0309-reasoning\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4.20-0309-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.5": {
"id": "xai/grok-4.5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/xai/grok-4.5\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-build-0.1": {
"id": "xai/grok-build-0.1",
"name": "Grok Build 0.1",
"description": "Fast Grok coding model tuned for agentic engineering and iterative edits",
"family": "grok-build",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 1,
"output": 2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/xai/grok-build-0.1\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-build-0.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.20-0309-non-reasoning": {
"id": "xai/grok-4.20-0309-non-reasoning",
"name": "Grok 4.20 (Non-Reasoning)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/xai/grok-4.20-0309-non-reasoning\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4.20-0309-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.7": {
"id": "zai/glm-4.7",
"name": "GLM-4.7",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/zai/glm-4.7\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.5-air": {
"id": "zai/glm-4.5-air",
"name": "GLM-4.5-Air",
"description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
"family": "glm-air",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0.2,
"output": 1.1,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/zai/glm-4.5-air\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.5-air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.6": {
"id": "zai/glm-4.6",
"name": "GLM-4.6",
"description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/zai/glm-4.6\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.2": {
"id": "zai/glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/zai/glm-5.2\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.5": {
"id": "zai/glm-4.5",
"name": "GLM-4.5",
"description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/zai/glm-4.5\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5": {
"id": "zai/glm-5",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/zai/glm-5\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.1": {
"id": "zai/glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/zai/glm-5.1\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5-turbo": {
"id": "zai/glm-5-turbo",
"name": "GLM-5-Turbo",
"description": "Faster GLM-5 lane for coding agents that need lower latency",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/zai/glm-5-turbo\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fireworks/glm-5.2": {
"id": "fireworks/glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/fireworks/glm-5.2\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"fireworks/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fireworks/gpt-oss-20b": {
"id": "fireworks/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.07,
"output": 0.3,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/fireworks/gpt-oss-20b\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"fireworks/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fireworks/gpt-oss-120b": {
"id": "fireworks/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/fireworks/gpt-oss-120b\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"fireworks/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras/gpt-oss-120b": {
"id": "cerebras/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.35,
"output": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"impossibl/cerebras/gpt-oss-120b\", apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.impossibl.com/v1\")!,\n apiKey: processEnvironment[\"IMPOSSIBL_API_KEY\"]\n)\nlet session = provider.model(\"cerebras/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"llmgateway-providers": {
"id": "llmgateway-providers",
"name": "LLM Gateway",
"baseURL": "https://api.llmgateway.io/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"LLMGATEWAY_API_KEY"
],
"doc": "https://llmgateway.io/docs",
"modelCount": 377,
"models": {
"vertex-openai/glm-4.7": {
"id": "vertex-openai/glm-4.7",
"name": "GLM-4.7 (Vertex AI (OpenAI-compatible))",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 128000
},
"cost": {
"input": 0.6,
"output": 2.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-openai/glm-4.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-openai/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-openai/qwen3-next-80b-a3b-thinking": {
"id": "vertex-openai/qwen3-next-80b-a3b-thinking",
"name": "Qwen3 Next 80B A3B Thinking (Vertex AI (OpenAI-compatible))",
"description": "Efficient Qwen thinking model for local reasoning, math, and coding agents",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-openai/qwen3-next-80b-a3b-thinking\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-openai/qwen3-next-80b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-openai/qwen3-next-80b-a3b-instruct": {
"id": "vertex-openai/qwen3-next-80b-a3b-instruct",
"name": "Qwen3 Next 80B A3B Instruct (Vertex AI (OpenAI-compatible))",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-openai/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-openai/qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-openai/kimi-k2-thinking": {
"id": "vertex-openai/kimi-k2-thinking",
"name": "Kimi K2 Thinking (Vertex AI (OpenAI-compatible))",
"description": "Thinking Kimi model for slower research passes, planning, and hard technical questions",
"family": "kimi-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-11-06",
"last_updated": "2025-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.6,
"output": 2.5,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-openai/kimi-k2-thinking\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-openai/kimi-k2-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-openai/deepseek-v3.2": {
"id": "vertex-openai/deepseek-v3.2",
"name": "DeepSeek V3.2 (Vertex AI (OpenAI-compatible))",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 65536
},
"cost": {
"input": 0.56,
"output": 1.68,
"cache_read": 0.056
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-openai/deepseek-v3.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-openai/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-openai/glm-5": {
"id": "vertex-openai/glm-5",
"name": "GLM-5 (Vertex AI (OpenAI-compatible))",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 32768
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-openai/glm-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-openai/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-openai/qwen3-235b-a22b-instruct-2507": {
"id": "vertex-openai/qwen3-235b-a22b-instruct-2507",
"name": "Qwen3 235B A22B Instruct 2507 (Vertex AI (OpenAI-compatible))",
"description": "Updated large open Qwen3 MoE instruct model for multilingual chat, coding, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-21",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.22,
"output": 0.88
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-openai/qwen3-235b-a22b-instruct-2507\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-openai/qwen3-235b-a22b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-openai/grok-4-6": {
"id": "vertex-openai/grok-4-6",
"name": "Grok 4.6 (Vertex AI (OpenAI-compatible))",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-openai/grok-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-openai/grok-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-openai/qwen3-coder-480b-a35b-instruct": {
"id": "vertex-openai/qwen3-coder-480b-a35b-instruct",
"name": "Qwen3 Coder 480B A35B Instruct (Vertex AI (OpenAI-compatible))",
"description": "Open Qwen coding heavyweight for repository reasoning and agentic engineering",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.22,
"output": 1.8,
"cache_read": 0.022
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-openai/qwen3-coder-480b-a35b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-openai/qwen3-coder-480b-a35b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-openai/grok-4-20-non-reasoning": {
"id": "vertex-openai/grok-4-20-non-reasoning",
"name": "Grok 4.20 Non-Reasoning (Vertex AI (OpenAI-compatible))",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-openai/grok-4-20-non-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-openai/grok-4-20-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-openai/grok-4-20-reasoning": {
"id": "vertex-openai/grok-4-20-reasoning",
"name": "Grok 4.20 Reasoning (Vertex AI (OpenAI-compatible))",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-openai/grok-4-20-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-openai/grok-4-20-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/kimi-k2.6": {
"id": "baidu/kimi-k2.6",
"name": "Kimi K2.6 (Baidu)",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/baidu/kimi-k2.6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"baidu/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/glm-5.2": {
"id": "baidu/glm-5.2",
"name": "GLM-5.2 (Baidu)",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/baidu/glm-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"baidu/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/deepseek-v4-flash": {
"id": "baidu/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (Baidu)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.44,
"output": 1.32,
"cache_read": 0.044
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/baidu/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"baidu/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/glm-5": {
"id": "baidu/glm-5",
"name": "GLM-5 (Baidu)",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/baidu/glm-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"baidu/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/glm-5.1": {
"id": "baidu/glm-5.1",
"name": "GLM-5.1 (Baidu)",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/baidu/glm-5.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"baidu/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/deepseek-v4-pro": {
"id": "baidu/deepseek-v4-pro",
"name": "DeepSeek V4 Pro (Baidu)",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.32,
"output": 3.96,
"cache_read": 0.132
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/baidu/deepseek-v4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"baidu/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/glm-5.3": {
"id": "baidu/glm-5.3",
"name": "GLM-5.3 (Baidu)",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/baidu/glm-5.3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"baidu/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-mantle/gpt-5.6-sol": {
"id": "aws-mantle/gpt-5.6-sol",
"name": "GPT-5.6 Sol (AWS Mantle)",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 278528,
"output": 128000
},
"cost": {
"input": 5.5,
"output": 33,
"cache_read": 0.55,
"cache_write": 6.875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-mantle/gpt-5.6-sol\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-mantle/gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-mantle/gpt-5.6-luna": {
"id": "aws-mantle/gpt-5.6-luna",
"name": "GPT-5.6 Luna (AWS Mantle)",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 278528,
"output": 128000
},
"cost": {
"input": 0.22,
"output": 1.32,
"cache_read": 0.022,
"cache_write": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-mantle/gpt-5.6-luna\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-mantle/gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-mantle/gpt-5.6-terra": {
"id": "aws-mantle/gpt-5.6-terra",
"name": "GPT-5.6 Terra (AWS Mantle)",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 278528,
"output": 128000
},
"cost": {
"input": 2.2,
"output": 13.2,
"cache_read": 0.22,
"cache_write": 2.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-mantle/gpt-5.6-terra\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-mantle/gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gonka24/minimax-m2.7": {
"id": "gonka24/minimax-m2.7",
"name": "MiniMax M2.7 (Gonka24)",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131100
},
"cost": {
"input": 0.08,
"output": 0.32,
"cache_read": 0.017
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/gonka24/minimax-m2.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gonka24/minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gonka24/deepseek-v4-flash": {
"id": "gonka24/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (Gonka24)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 390000,
"output": 16384
},
"cost": {
"input": 0.051,
"output": 0.104,
"cache_read": 0.0097
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/gonka24/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gonka24/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"embercloud/glm-4.7": {
"id": "embercloud/glm-4.7",
"name": "GLM-4.7 (EmberCloud)",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131000
},
"cost": {
"input": 0.38,
"output": 1.98,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/embercloud/glm-4.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"embercloud/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"embercloud/glm-4.5-air": {
"id": "embercloud/glm-4.5-air",
"name": "GLM-4.5 Air (EmberCloud)",
"description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
"family": "glm-air",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131000,
"output": 96000
},
"cost": {
"input": 0.13,
"output": 0.85,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/embercloud/glm-4.5-air\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"embercloud/glm-4.5-air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"embercloud/glm-5.2": {
"id": "embercloud/glm-5.2",
"name": "GLM-5.2 (EmberCloud)",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 203000,
"output": 131000
},
"cost": {
"input": 1.26,
"output": 3.96,
"cache_read": 0.234
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/embercloud/glm-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"embercloud/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"embercloud/qwen3-coder-next": {
"id": "embercloud/qwen3-coder-next",
"name": "Qwen3 Coder Next (EmberCloud)",
"description": "Open-weight Qwen coding model for agents, repository edits, and multi-turn tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2026-02-03",
"last_updated": "2026-02-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.108,
"output": 0.675,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/embercloud/qwen3-coder-next\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"embercloud/qwen3-coder-next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"embercloud/glm-4.5": {
"id": "embercloud/glm-4.5",
"name": "GLM-4.5 (EmberCloud)",
"description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131000,
"output": 96000
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/embercloud/glm-4.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"embercloud/glm-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"embercloud/glm-5": {
"id": "embercloud/glm-5",
"name": "GLM-5 (EmberCloud)",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 203000,
"output": 131000
},
"cost": {
"input": 0.72,
"output": 2.3,
"cache_read": 0.144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/embercloud/glm-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"embercloud/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"embercloud/kimi-k2.5": {
"id": "embercloud/kimi-k2.5",
"name": "Kimi K2.5 (EmberCloud)",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.405,
"output": 1.98,
"cache_read": 0.225
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/embercloud/kimi-k2.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"embercloud/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"embercloud/glm-5.1": {
"id": "embercloud/glm-5.1",
"name": "GLM-5.1 (EmberCloud)",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 203000,
"output": 131000
},
"cost": {
"input": 0.931,
"output": 2.93,
"cache_read": 0.173
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/embercloud/glm-5.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"embercloud/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"embercloud/glm-4.7-flash": {
"id": "embercloud/glm-4.7-flash",
"name": "GLM-4.7 Flash (EmberCloud)",
"description": "Budget GLM lane for fast coding help, routing, and everyday automation",
"family": "glm-flash",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131000
},
"cost": {
"input": 0.06,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/embercloud/glm-4.7-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"embercloud/glm-4.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai/minimax-m2.7": {
"id": "scx-ai/minimax-m2.7",
"name": "MiniMax M2.7 (SCX.ai (Turbo))",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 196608
},
"cost": {
"input": 0.48,
"output": 1.79,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai/minimax-m2.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai/minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai/qwen3-32b": {
"id": "scx-ai/qwen3-32b",
"name": "Qwen3 32B (SCX.ai (Turbo))",
"description": "Dense open Qwen model for self-hosted chat, reasoning, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0.36,
"output": 0.87
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai/qwen3-32b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai/qwen3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai/llama-4-maverick-17b-instruct": {
"id": "scx-ai/llama-4-maverick-17b-instruct",
"name": "Llama 4 Maverick 17B Instruct (SCX.ai (Turbo))",
"description": "Open multimodal Llama for strong reasoning with efficient everyday serving",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.53,
"output": 1.62
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai/llama-4-maverick-17b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai/llama-4-maverick-17b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai/gemma-4-31b-it": {
"id": "scx-ai/gemma-4-31b-it",
"name": "Gemma 4 31B IT (SCX.ai (Turbo))",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.3,
"output": 0.91
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai/gemma-4-31b-it\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai/gpt-oss-120b": {
"id": "scx-ai/gpt-oss-120b",
"name": "GPT OSS 120B (SCX.ai (Turbo))",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.17,
"output": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai/gpt-oss-120b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"groq/gpt-oss-20b": {
"id": "groq/gpt-oss-20b",
"name": "GPT OSS 20B (Groq)",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32766
},
"cost": {
"input": 0.1,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/groq/gpt-oss-20b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"groq/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"groq/gpt-oss-120b": {
"id": "groq/gpt-oss-120b",
"name": "GPT OSS 120B (Groq)",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32766
},
"cost": {
"input": 0.15,
"output": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/groq/gpt-oss-120b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"groq/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-vertex/gemini-3.1-pro-preview": {
"id": "google-vertex/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro (Preview) (Google Vertex AI)",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-vertex/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-vertex/gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-vertex/gemini-2.5-flash-lite": {
"id": "google-vertex/gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash Lite (Google Vertex AI)",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-vertex/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-vertex/gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-vertex/gemini-3.6-flash": {
"id": "google-vertex/gemini-3.6-flash",
"name": "Gemini 3.6 Flash (Google Vertex AI)",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-vertex/gemini-3.6-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-vertex/gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-vertex/gemini-3.1-flash-lite": {
"id": "google-vertex/gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite (Google Vertex AI)",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-vertex/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-vertex/gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-vertex/gemini-3.5-flash": {
"id": "google-vertex/gemini-3.5-flash",
"name": "Gemini 3.5 Flash (Google Vertex AI)",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-vertex/gemini-3.5-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-vertex/gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-vertex/gemini-3.5-flash-lite": {
"id": "google-vertex/gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite (Google Vertex AI)",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-vertex/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-vertex/gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-vertex/gemini-3-flash-preview": {
"id": "google-vertex/gemini-3-flash-preview",
"name": "Gemini 3 Flash (Preview) (Google Vertex AI)",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-vertex/gemini-3-flash-preview\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-vertex/gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-vertex/gemini-3.8-flash": {
"id": "google-vertex/gemini-3.8-flash",
"name": "Gemini 3.8 Flash (Google Vertex AI)",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-vertex/gemini-3.8-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-vertex/gemini-3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-vertex/gemini-3.7-flash": {
"id": "google-vertex/gemini-3.7-flash",
"name": "Gemini 3.7 Flash (Google Vertex AI)",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-vertex/gemini-3.7-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-vertex/gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-vertex/gemini-2.5-pro": {
"id": "google-vertex/gemini-2.5-pro",
"name": "Gemini 2.5 Pro (Google Vertex AI)",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-vertex/gemini-2.5-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-vertex/gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-vertex/gemini-2.5-flash": {
"id": "google-vertex/gemini-2.5-flash",
"name": "Gemini 2.5 Flash (Google Vertex AI)",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1,
"max": 24576
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-vertex/gemini-2.5-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-vertex/gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"quartz/gemini-3.1-pro-preview": {
"id": "quartz/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro (Preview) (Quartz)",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/quartz/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"quartz/gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-anthropic/claude-sonnet-4-6": {
"id": "vertex-anthropic/claude-sonnet-4-6",
"name": "Claude Sonnet 4.6 (Vertex AI (Anthropic))",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-anthropic/claude-sonnet-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-anthropic/claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-anthropic/claude-opus-4-6": {
"id": "vertex-anthropic/claude-opus-4-6",
"name": "Claude Opus 4.6 (Vertex AI (Anthropic))",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-anthropic/claude-opus-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-anthropic/claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-anthropic/claude-opus-4-7": {
"id": "vertex-anthropic/claude-opus-4-7",
"name": "Claude Opus 4.7 (Vertex AI (Anthropic))",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-anthropic/claude-opus-4-7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-anthropic/claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-anthropic/claude-haiku-4-5": {
"id": "vertex-anthropic/claude-haiku-4-5",
"name": "Claude Haiku 4.5 (Vertex AI (Anthropic))",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-anthropic/claude-haiku-4-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-anthropic/claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-anthropic/claude-sonnet-4-5": {
"id": "vertex-anthropic/claude-sonnet-4-5",
"name": "Claude Sonnet 4.5 (Vertex AI (Anthropic))",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-anthropic/claude-sonnet-4-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-anthropic/claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-anthropic/claude-sonnet-5": {
"id": "vertex-anthropic/claude-sonnet-5",
"name": "Claude Sonnet 5 (Vertex AI (Anthropic))",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-anthropic/claude-sonnet-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-anthropic/claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertex-anthropic/claude-opus-4-5-20251101": {
"id": "vertex-anthropic/claude-opus-4-5-20251101",
"name": "Claude Opus 4.5 (Vertex AI (Anthropic))",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 31999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-01",
"last_updated": "2025-11-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vertex-anthropic/claude-opus-4-5-20251101\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vertex-anthropic/claude-opus-4-5-20251101\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5": {
"id": "xiaomi/mimo-v2.5",
"name": "MiMo V2.5 (Xiaomi)",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028,
"tiers": [
{
"input": 0.8,
"output": 4,
"cache_read": 0.16,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 0.8,
"output": 4,
"cache_read": 0.16
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/xiaomi/mimo-v2.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5-pro": {
"id": "xiaomi/mimo-v2.5-pro",
"name": "MiMo V2.5 Pro (Xiaomi)",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.0036,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/xiaomi/mimo-v2.5-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.1-lightning": {
"id": "minimax/minimax-m2.1-lightning",
"name": "MiniMax M2.1 Lightning (MiniMax)",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 196608,
"output": 131072
},
"cost": {
"input": 0.12,
"output": 0.48
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/minimax/minimax-m2.1-lightning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.1-lightning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.1": {
"id": "minimax/minimax-m2.1",
"name": "MiniMax M2.1 (MiniMax)",
"description": "Earlier MiniMax agent model for practical coding and productivity tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 131072
},
"cost": {
"input": 0.27,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/minimax/minimax-m2.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2": {
"id": "minimax/minimax-m2",
"name": "MiniMax M2 (MiniMax)",
"description": "Efficient open MiniMax model built for coding agents and tool-heavy workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-10-27",
"last_updated": "2025-10-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 131072
},
"cost": {
"input": 0.2,
"output": 1,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/minimax/minimax-m2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7-highspeed": {
"id": "minimax/minimax-m2.7-highspeed",
"name": "MiniMax M2.7 Highspeed (MiniMax)",
"description": "Low-latency M2.7 variant for interactive coding plans and agent loops",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131100
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/minimax/minimax-m2.7-highspeed\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7": {
"id": "minimax/minimax-m2.7",
"name": "MiniMax M2.7 (MiniMax)",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131100
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/minimax/minimax-m2.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5": {
"id": "minimax/minimax-m2.5",
"name": "MiniMax M2.5 (MiniMax)",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131100
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/minimax/minimax-m2.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m3": {
"id": "minimax/minimax-m3",
"name": "MiniMax M3 (MiniMax)",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512000,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/minimax/minimax-m3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5-highspeed": {
"id": "minimax/minimax-m2.5-highspeed",
"name": "MiniMax M2.5 Highspeed (MiniMax)",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-13",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131100
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/minimax/minimax-m2.5-highspeed\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-text-01": {
"id": "minimax/minimax-text-01",
"name": "MiniMax Text 01 (MiniMax)",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-01-15",
"last_updated": "2025-01-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.2,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/minimax/minimax-text-01\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-text-01\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.7-max": {
"id": "alibaba/qwen3.7-max",
"name": "Qwen3.7 Max (Alibaba Cloud)",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.5,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3.7-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-coder-plus": {
"id": "alibaba/qwen3-coder-plus",
"name": "Qwen3 Coder Plus (Alibaba Cloud)",
"description": "Hosted Qwen coder for software agents, repo edits, and long-context code",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 66000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.2,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3-coder-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen35-397b-a17b": {
"id": "alibaba/qwen35-397b-a17b",
"name": "Qwen3.5 397B A17B (Alibaba Cloud)",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen35-397b-a17b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen35-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-coder-flash": {
"id": "alibaba/qwen3-coder-flash",
"name": "Qwen3 Coder Flash (Alibaba Cloud)",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.5,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3-coder-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-coder-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen-max": {
"id": "alibaba/qwen-max",
"name": "Qwen Max (Alibaba Cloud)",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-04-03",
"last_updated": "2025-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 1.6,
"output": 6.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.6-plus": {
"id": "alibaba/qwen3.6-plus",
"name": "Qwen3.6 Plus (Alibaba Cloud)",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"cache_write": 0.625,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3.6-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen-flash": {
"id": "alibaba/qwen-flash",
"name": "Qwen Flash (Alibaba Cloud)",
"description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.01,
"cache_write": 0.0625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/glm-5.2": {
"id": "alibaba/glm-5.2",
"name": "GLM-5.2 (Alibaba Cloud)",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/glm-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/deepseek-v4-flash": {
"id": "alibaba/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (Alibaba Cloud)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 393216
},
"cost": {
"input": 0.2,
"output": 0.4,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-vl-plus": {
"id": "alibaba/qwen3-vl-plus",
"name": "Qwen3 VL Plus (Alibaba Cloud)",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 1.6,
"cache_read": 0.04,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3-vl-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-vl-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen-coder-plus": {
"id": "alibaba/qwen-coder-plus",
"name": "Qwen Coder Plus (Alibaba Cloud)",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2024-09-18",
"last_updated": "2024-09-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.502,
"output": 1.004
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen-coder-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.7-flash": {
"id": "alibaba/qwen3.7-flash",
"name": "Qwen3.7 Flash (Alibaba Cloud)",
"description": "Lightweight multimodal Qwen model for high-throughput text, image, and video tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 991000,
"output": 65536
},
"cost": {
"input": 0.03,
"output": 0.13,
"cache_read": 0.006,
"cache_write": 0.0375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3.7-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.6-35b-a3b": {
"id": "alibaba/qwen3.6-35b-a3b",
"name": "Qwen3.6 35B A3B (Alibaba Cloud)",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.375,
"output": 2.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3.6-35b-a3b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-max": {
"id": "alibaba/qwen3-max",
"name": "Qwen3 Max (Alibaba Cloud)",
"description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32800
},
"cost": {
"input": 1.2,
"output": 6,
"cache_read": 0.24,
"cache_write": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-vl-flash": {
"id": "alibaba/qwen3-vl-flash",
"name": "Qwen3 VL Flash (Alibaba Cloud)",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3-vl-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-vl-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen-plus": {
"id": "alibaba/qwen-plus",
"name": "Qwen Plus (Alibaba Cloud)",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01-25",
"last_updated": "2025-09-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32000
},
"cost": {
"input": 0.4,
"output": 1.2,
"cache_read": 0.08,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.6-flash": {
"id": "alibaba/qwen3.6-flash",
"name": "Qwen3.6 Flash (Alibaba Cloud)",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.05,
"cache_write": 0.3125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3.6-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.8-flash": {
"id": "alibaba/qwen3.8-flash",
"name": "Qwen3.8 Flash (Alibaba Cloud)",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.47,
"cache_read": 0.016,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3.8-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.6-max-preview": {
"id": "alibaba/qwen3.6-max-preview",
"name": "Qwen3.6 Max Preview (Alibaba Cloud)",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-20",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 1.3,
"output": 7.8,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3.6-max-preview\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.6-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/glm-5": {
"id": "alibaba/glm-5",
"name": "GLM-5 (Alibaba Cloud)",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 16384
},
"cost": {
"input": 0.573,
"output": 2.58
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/glm-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.8-max": {
"id": "alibaba/qwen3.8-max",
"name": "Qwen3.8 Max (Alibaba Cloud)",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3.8-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/kimi-k2.5": {
"id": "alibaba/kimi-k2.5",
"name": "Kimi K2.5 (Alibaba Cloud)",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 98304
},
"cost": {
"input": 0.574,
"output": 3.011
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/kimi-k2.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.7-plus": {
"id": "alibaba/qwen3.7-plus",
"name": "Qwen3.7 Plus (Alibaba Cloud)",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.08,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen3.7-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen-omni-turbo": {
"id": "alibaba/qwen-omni-turbo",
"name": "Qwen Omni Turbo (Alibaba Cloud)",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-01-19",
"last_updated": "2025-03-26",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0.2,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen-omni-turbo\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen-omni-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/deepseek-v4-pro": {
"id": "alibaba/deepseek-v4-pro",
"name": "DeepSeek V4 Pro (Alibaba Cloud)",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 393216
},
"cost": {
"input": 2.4,
"output": 4.8,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/deepseek-v4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen-plus-latest": {
"id": "alibaba/qwen-plus-latest",
"name": "Qwen Plus Latest (Alibaba Cloud)",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2024-09-09",
"last_updated": "2024-09-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32000
},
"cost": {
"input": 0.4,
"output": 1.2,
"cache_read": 0.08,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/alibaba/qwen-plus-latest\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen-plus-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai-gp/glm-5.2": {
"id": "scx-ai-gp/glm-5.2",
"name": "GLM-5.2 (SCX.ai)",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.8,
"output": 2.55,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai-gp/glm-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai-gp/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai-gp/kimi-k2.7-code": {
"id": "scx-ai-gp/kimi-k2.7-code",
"name": "Kimi K2.7 Code (SCX.ai)",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai-gp/kimi-k2.7-code\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai-gp/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai-gp/kimi-k3": {
"id": "scx-ai-gp/kimi-k3",
"name": "Kimi K3 (SCX.ai)",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 3.5,
"output": 18,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai-gp/kimi-k3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai-gp/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai-gp/glm-5.2-fast": {
"id": "scx-ai-gp/glm-5.2-fast",
"name": "GLM-5.2 Turbo (SCX.ai)",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2.2,
"output": 6.5,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai-gp/glm-5.2-fast\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai-gp/glm-5.2-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai-gp/glm-5.3-flash": {
"id": "scx-ai-gp/glm-5.3-flash",
"name": "GLM-5.3 Flash (SCX.ai)",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.088,
"output": 0.25,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai-gp/glm-5.3-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai-gp/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai-gp/qwen3.8-max": {
"id": "scx-ai-gp/qwen3.8-max",
"name": "Qwen3.8 Max (SCX.ai)",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai-gp/qwen3.8-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai-gp/qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"scx-ai-gp/glm-5.3": {
"id": "scx-ai-gp/glm-5.3",
"name": "GLM-5.3 (SCX.ai)",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/scx-ai-gp/glm-5.3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"scx-ai-gp/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-sonnet-4-6": {
"id": "aws-bedrock/claude-sonnet-4-6",
"name": "Claude Sonnet 4.6 (AWS Bedrock)",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-sonnet-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/llama-4-scout-17b-instruct": {
"id": "aws-bedrock/llama-4-scout-17b-instruct",
"name": "Llama 4 Scout 17B Instruct (AWS Bedrock)",
"description": "Open Llama with long-context vision for efficient multimodal agents",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 2048
},
"cost": {
"input": 0.17,
"output": 0.66
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/llama-4-scout-17b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/llama-4-scout-17b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-opus-5": {
"id": "aws-bedrock/claude-opus-5",
"name": "Claude Opus 5 (AWS Bedrock)",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-opus-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-opus-4-1-20250805": {
"id": "aws-bedrock/claude-opus-4-1-20250805",
"name": "Claude Opus 4.1 (AWS Bedrock)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 31999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-opus-4-1-20250805\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-opus-4-1-20250805\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-fable-5-1": {
"id": "aws-bedrock/claude-fable-5-1",
"name": "Claude Fable 5.1 (AWS Bedrock)",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-fable-5-1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-fable-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-opus-4-6": {
"id": "aws-bedrock/claude-opus-4-6",
"name": "Claude Opus 4.6 (AWS Bedrock)",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-opus-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-sonnet-4-5-20250929": {
"id": "aws-bedrock/claude-sonnet-4-5-20250929",
"name": "Claude Sonnet 4.5 (2025-09-29) (AWS Bedrock)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-sonnet-4-5-20250929\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-sonnet-4-5-20250929\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-opus-4-7": {
"id": "aws-bedrock/claude-opus-4-7",
"name": "Claude Opus 4.7 (AWS Bedrock)",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-opus-4-7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-haiku-4-5-20251001": {
"id": "aws-bedrock/claude-haiku-4-5-20251001",
"name": "Claude Haiku 4.5 (2025-10-01) (AWS Bedrock)",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-haiku-4-5-20251001\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-haiku-4-5-20251001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-fable-5": {
"id": "aws-bedrock/claude-fable-5",
"name": "Claude Fable 5 (AWS Bedrock)",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-fable-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/llama-4-maverick-17b-instruct": {
"id": "aws-bedrock/llama-4-maverick-17b-instruct",
"name": "Llama 4 Maverick 17B Instruct (AWS Bedrock)",
"description": "Open multimodal Llama for strong reasoning with efficient everyday serving",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 2048
},
"cost": {
"input": 0.24,
"output": 0.97
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/llama-4-maverick-17b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/llama-4-maverick-17b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/grok-4-3": {
"id": "aws-bedrock/grok-4-3",
"name": "Grok 4.3 (AWS Bedrock)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/grok-4-3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/grok-4-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-haiku-4-5": {
"id": "aws-bedrock/claude-haiku-4-5",
"name": "Claude Haiku 4.5 (AWS Bedrock)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-haiku-4-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-sonnet-4-5": {
"id": "aws-bedrock/claude-sonnet-4-5",
"name": "Claude Sonnet 4.5 (AWS Bedrock)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-sonnet-4-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/llama-3.1-70b-instruct": {
"id": "aws-bedrock/llama-3.1-70b-instruct",
"name": "Llama 3.1 70B Instruct (AWS Bedrock)",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 2048
},
"cost": {
"input": 0.72,
"output": 0.72
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/llama-3.1-70b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/llama-3.1-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/grok-4-6": {
"id": "aws-bedrock/grok-4-6",
"name": "Grok 4.6 (AWS Bedrock)",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/grok-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/grok-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-opus-4-8": {
"id": "aws-bedrock/claude-opus-4-8",
"name": "Claude Opus 4.8 (AWS Bedrock)",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-opus-4-8\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-sonnet-5": {
"id": "aws-bedrock/claude-sonnet-5",
"name": "Claude Sonnet 5 (AWS Bedrock)",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-sonnet-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aws-bedrock/claude-opus-4-5-20251101": {
"id": "aws-bedrock/claude-opus-4-5-20251101",
"name": "Claude Opus 4.5 (AWS Bedrock)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 31999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-01",
"last_updated": "2025-11-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/aws-bedrock/claude-opus-4-5-20251101\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"aws-bedrock/claude-opus-4-5-20251101\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4-6": {
"id": "anthropic/claude-sonnet-4-6",
"name": "Claude Sonnet 4.6 (Anthropic)",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-sonnet-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-5": {
"id": "anthropic/claude-opus-5",
"name": "Claude Opus 5 (Anthropic)",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-opus-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5-1": {
"id": "anthropic/claude-fable-5-1",
"name": "Claude Fable 5.1 (Anthropic)",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-fable-5-1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-6": {
"id": "anthropic/claude-opus-4-6",
"name": "Claude Opus 4.6 (Anthropic)",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-opus-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4-5-20250929": {
"id": "anthropic/claude-sonnet-4-5-20250929",
"name": "Claude Sonnet 4.5 (2025-09-29) (Anthropic)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-sonnet-4-5-20250929\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4-5-20250929\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-7": {
"id": "anthropic/claude-opus-4-7",
"name": "Claude Opus 4.7 (Anthropic)",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-opus-4-7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-4-5-20251001": {
"id": "anthropic/claude-haiku-4-5-20251001",
"name": "Claude Haiku 4.5 (2025-10-01) (Anthropic)",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-haiku-4-5-20251001\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-4-5-20251001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5": {
"id": "anthropic/claude-fable-5",
"name": "Claude Fable 5 (Anthropic)",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-fable-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-4-5": {
"id": "anthropic/claude-haiku-4-5",
"name": "Claude Haiku 4.5 (Anthropic)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-haiku-4-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4-5": {
"id": "anthropic/claude-sonnet-4-5",
"name": "Claude Sonnet 4.5 (Anthropic)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-sonnet-4-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-8": {
"id": "anthropic/claude-opus-4-8",
"name": "Claude Opus 4.8 (Anthropic)",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-opus-4-8\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-5": {
"id": "anthropic/claude-sonnet-5",
"name": "Claude Sonnet 5 (Anthropic)",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-sonnet-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-5-20251101": {
"id": "anthropic/claude-opus-4-5-20251101",
"name": "Claude Opus 4.5 (Anthropic)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 31999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-01",
"last_updated": "2025-11-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/anthropic/claude-opus-4-5-20251101\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-5-20251101\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"canopywave/kimi-k2.6": {
"id": "canopywave/kimi-k2.6",
"name": "Kimi K2.6 (CanopyWave)",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/canopywave/kimi-k2.6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"canopywave/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"canopywave/glm-5.2": {
"id": "canopywave/glm-5.2",
"name": "GLM-5.2 (CanopyWave)",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 32768
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/canopywave/glm-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"canopywave/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"canopywave/deepseek-v4-flash": {
"id": "canopywave/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (CanopyWave)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 393216
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/canopywave/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"canopywave/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"canopywave/kimi-k3": {
"id": "canopywave/kimi-k3",
"name": "Kimi K3 (CanopyWave)",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/canopywave/kimi-k3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"canopywave/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"canopywave/deepseek-v4-pro": {
"id": "canopywave/deepseek-v4-pro",
"name": "DeepSeek V4 Pro (CanopyWave)",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 393216
},
"cost": {
"input": 1.74,
"output": 3.48,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/canopywave/deepseek-v4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"canopywave/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vichar-ai/glm-5.3-flash": {
"id": "vichar-ai/glm-5.3-flash",
"name": "GLM-5.3 Flash (vichar-ai)",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048000,
"output": 128000
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vichar-ai/glm-5.3-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vichar-ai/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vichar-ai/glm-5.3": {
"id": "vichar-ai/glm-5.3",
"name": "GLM-5.3 (vichar-ai)",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048000,
"output": 128000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/vichar-ai/glm-5.3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"vichar-ai/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"together-ai/glm-4.7": {
"id": "together-ai/glm-4.7",
"name": "GLM-4.7 (Together AI)",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": false,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 128000
},
"cost": {
"input": 0.45,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/together-ai/glm-4.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"together-ai/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"together-ai/minimax-m3": {
"id": "together-ai/minimax-m3",
"name": "MiniMax M3 (Together AI)",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/together-ai/minimax-m3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"together-ai/minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"together-ai/deepseek-v4-flash": {
"id": "together-ai/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (Together AI)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/together-ai/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"together-ai/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"together-ai/gpt-oss-20b": {
"id": "together-ai/gpt-oss-20b",
"name": "GPT OSS 20B (Together AI)",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/together-ai/gpt-oss-20b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"together-ai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"together-ai/kimi-k3": {
"id": "together-ai/kimi-k3",
"name": "Kimi K3 (Together AI)",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/together-ai/kimi-k3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"together-ai/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"together-ai/gemma-4-31b-it": {
"id": "together-ai/gemma-4-31b-it",
"name": "Gemma 4 31B IT (Together AI)",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.39,
"output": 0.97
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/together-ai/gemma-4-31b-it\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"together-ai/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"together-ai/deepseek-v4-pro": {
"id": "together-ai/deepseek-v4-pro",
"name": "DeepSeek V4 Pro (Together AI)",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 163840
},
"cost": {
"input": 1.32,
"output": 3.96,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/together-ai/deepseek-v4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"together-ai/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"together-ai/gpt-oss-120b": {
"id": "together-ai/gpt-oss-120b",
"name": "GPT OSS 120B (Together AI)",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/together-ai/gpt-oss-120b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"together-ai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.3": {
"id": "meta/muse-spark-1.3",
"name": "Muse Spark 1.3 (Meta)",
"description": "Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It improves long-horizon agent collaboration, instruction following, and coding efficiency relative to Muse Spark 1.2.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/meta/muse-spark-1.3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.2": {
"id": "meta/muse-spark-1.2",
"name": "Muse Spark 1.2 (Meta)",
"description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/meta/muse-spark-1.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.1": {
"id": "meta/muse-spark-1.1",
"name": "Muse Spark 1.1 (Meta)",
"description": "Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-08",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/meta/muse-spark-1.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-pro-latest": {
"id": "google-ai-studio/gemini-pro-latest",
"name": "Gemini Pro Latest (Google AI Studio)",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-27",
"last_updated": "2026-02-27",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-pro-latest\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-pro-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-3.1-pro-preview": {
"id": "google-ai-studio/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro (Preview) (Google AI Studio)",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-2.5-flash-lite": {
"id": "google-ai-studio/gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash Lite (Google AI Studio)",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-3.6-flash": {
"id": "google-ai-studio/gemini-3.6-flash",
"name": "Gemini 3.6 Flash (Google AI Studio)",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-3.6-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-3.1-flash-lite": {
"id": "google-ai-studio/gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite (Google AI Studio)",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-3.5-flash": {
"id": "google-ai-studio/gemini-3.5-flash",
"name": "Gemini 3.5 Flash (Google AI Studio)",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-3.5-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-3.5-flash-lite": {
"id": "google-ai-studio/gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite (Google AI Studio)",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-3-flash-preview": {
"id": "google-ai-studio/gemini-3-flash-preview",
"name": "Gemini 3 Flash (Preview) (Google AI Studio)",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-3-flash-preview\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-3.8-flash": {
"id": "google-ai-studio/gemini-3.8-flash",
"name": "Gemini 3.8 Flash (Google AI Studio)",
"description": "Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-3.8-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-3.7-flash": {
"id": "google-ai-studio/gemini-3.7-flash",
"name": "Gemini 3.7 Flash (Google AI Studio)",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-3.7-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-2.5-pro": {
"id": "google-ai-studio/gemini-2.5-pro",
"name": "Gemini 2.5 Pro (Google AI Studio)",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-2.5-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-ai-studio/gemini-2.5-flash": {
"id": "google-ai-studio/gemini-2.5-flash",
"name": "Gemini 2.5 Flash (Google AI Studio)",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1,
"max": 24576
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/google-ai-studio/gemini-2.5-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google-ai-studio/gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/glm-4.7": {
"id": "bytedance/glm-4.7",
"name": "GLM-4.7 (ByteDance)",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/bytedance/glm-4.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seed-1-8-251228": {
"id": "bytedance/seed-1-8-251228",
"name": "Seed 1.8 (251228) (ByteDance)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-18",
"last_updated": "2025-12-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/bytedance/seed-1-8-251228\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seed-1-8-251228\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/glm-5.2": {
"id": "bytedance/glm-5.2",
"name": "GLM-5.2 (ByteDance)",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1024000,
"output": 128000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/bytedance/glm-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/deepseek-v4-flash": {
"id": "bytedance/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (ByteDance)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393216
},
"cost": {
"input": 0.44,
"output": 1.32,
"cache_read": 0.014
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/bytedance/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seed-1-6-flash-250715": {
"id": "bytedance/seed-1-6-flash-250715",
"name": "Seed 1.6 Flash (250715) (ByteDance)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-26",
"last_updated": "2025-07-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.07,
"output": 0.3,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/bytedance/seed-1-6-flash-250715\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seed-1-6-flash-250715\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/deepseek-v3.2": {
"id": "bytedance/deepseek-v3.2",
"name": "DeepSeek V3.2 (ByteDance)",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.28,
"output": 0.42,
"cache_read": 0.056
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/bytedance/deepseek-v3.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seed-1-6-250615": {
"id": "bytedance/seed-1-6-250615",
"name": "Seed 1.6 (250615) (ByteDance)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-06-25",
"last_updated": "2025-06-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/bytedance/seed-1-6-250615\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seed-1-6-250615\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/deepseek-v4-pro": {
"id": "bytedance/deepseek-v4-pro",
"name": "DeepSeek V4 Pro (ByteDance)",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393216
},
"cost": {
"input": 1.32,
"output": 3.96,
"cache_read": 0.044
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/bytedance/deepseek-v4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/gpt-oss-120b": {
"id": "bytedance/gpt-oss-120b",
"name": "GPT OSS 120B (ByteDance)",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32000
},
"cost": {
"input": 0.1,
"output": 0.5,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/bytedance/gpt-oss-120b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seed-1-6-250915": {
"id": "bytedance/seed-1-6-250915",
"name": "Seed 1.6 (250915) (ByteDance)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-09-15",
"last_updated": "2025-09-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/bytedance/seed-1-6-250915\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seed-1-6-250915\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-4.7": {
"id": "novita/glm-4.7",
"name": "GLM-4.7 (NovitaAI)",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 128000
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/glm-4.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3.7-max": {
"id": "novita/qwen3.7-max",
"name": "Qwen3.7 Max (NovitaAI)",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 3.75,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3.7-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/gemma-4-26b-a4b-it": {
"id": "novita/gemma-4-26b-a4b-it",
"name": "Gemma 4 26B A4B IT (NovitaAI)",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.13,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/llama-4-scout-17b-instruct": {
"id": "novita/llama-4-scout-17b-instruct",
"name": "Llama 4 Scout 17B Instruct (NovitaAI)",
"description": "Open Llama with long-context vision for efficient multimodal agents",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.18,
"output": 0.59
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/llama-4-scout-17b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/llama-4-scout-17b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen35-397b-a17b": {
"id": "novita/qwen35-397b-a17b",
"name": "Qwen3.5 397B A17B (NovitaAI)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 64000
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen35-397b-a17b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen35-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3-235b-a22b-thinking-2507": {
"id": "novita/qwen3-235b-a22b-thinking-2507",
"name": "Qwen3 235B A22B Thinking 2507 (NovitaAI)",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-25",
"last_updated": "2025-07-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3-235b-a22b-thinking-2507\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3-235b-a22b-thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-4.6": {
"id": "novita/glm-4.6",
"name": "GLM-4.6 (NovitaAI)",
"description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.55,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/glm-4.6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/minimax-m2.1": {
"id": "novita/minimax-m2.1",
"name": "MiniMax M2.1 (NovitaAI)",
"description": "Earlier MiniMax agent model for practical coding and productivity tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/minimax-m2.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/minimax-m2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-4.6v": {
"id": "novita/glm-4.6v",
"name": "GLM-4.6V (NovitaAI)",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16000
},
"cost": {
"input": 0.3,
"output": 0.9,
"cache_read": 0.055
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/glm-4.6v\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-4.6v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3-next-80b-a3b-instruct": {
"id": "novita/qwen3-next-80b-a3b-instruct",
"name": "Qwen3 Next 80B A3B Instruct (NovitaAI)",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/ling-3.0-flash": {
"id": "novita/ling-3.0-flash",
"name": "InclusionAI Ling 3.0 Flash (NovitaAI)",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "ling",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-02",
"last_updated": "2026-08-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.06,
"output": 0.18,
"cache_read": 0.012
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/ling-3.0-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/ling-3.0-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3.8-27b": {
"id": "novita/qwen3.8-27b",
"name": "Qwen3.8 27B (NovitaAI)",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.42,
"output": 3,
"cache_read": 0.085
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3.8-27b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3-235b-a22b-fp8": {
"id": "novita/qwen3-235b-a22b-fp8",
"name": "Qwen3 235B A22B FP8 (NovitaAI)",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-04-28",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 40960,
"output": 20000
},
"cost": {
"input": 0.2,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3-235b-a22b-fp8\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3-235b-a22b-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/minimax-m2.7": {
"id": "novita/minimax-m2.7",
"name": "MiniMax M2.7 (NovitaAI)",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131100
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/minimax-m2.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/kimi-k2.6": {
"id": "novita/kimi-k2.6",
"name": "Kimi K2.6 (NovitaAI)",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.8,
"output": 3.4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/kimi-k2.6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-5.2": {
"id": "novita/glm-5.2",
"name": "GLM-5.2 (NovitaAI)",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/glm-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/minimax-m2.5": {
"id": "novita/minimax-m2.5",
"name": "MiniMax M2.5 (NovitaAI)",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131100
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/minimax-m2.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/deepseek-v4-flash": {
"id": "novita/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (NovitaAI)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1050000,
"output": 393216
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/kimi-k2.7-code": {
"id": "novita/kimi-k2.7-code",
"name": "Kimi K2.7 Code (NovitaAI)",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/kimi-k2.7-code\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/llama-3.2-3b-instruct": {
"id": "novita/llama-3.2-3b-instruct",
"name": "Llama 3.2 3B Instruct (NovitaAI)",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2024-09-18",
"last_updated": "2024-09-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32000
},
"cost": {
"input": 0.03,
"output": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/llama-3.2-3b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/llama-3.2-3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/hy3": {
"id": "novita/hy3",
"name": "Hy3 (NovitaAI)",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 192000,
"output": 262144
},
"cost": {
"input": 0.14,
"output": 0.58,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/hy3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3-coder-30b-a3b-instruct": {
"id": "novita/qwen3-coder-30b-a3b-instruct",
"name": "Qwen3 Coder 30B A3B Instruct (NovitaAI)",
"description": "Smaller Qwen coder for efficient local agents and repo-level fixes",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 160000,
"output": 32768
},
"cost": {
"input": 0.07,
"output": 0.27
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/ernie-4.5-vl-424b-a47b": {
"id": "novita/ernie-4.5-vl-424b-a47b",
"name": "ERNIE 4.5 VL 424B A47B (NovitaAI)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "ernie",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-06-30",
"last_updated": "2025-06-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 123000,
"output": 16000
},
"cost": {
"input": 0.42,
"output": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/ernie-4.5-vl-424b-a47b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/ernie-4.5-vl-424b-a47b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/kimi-k3": {
"id": "novita/kimi-k3",
"name": "Kimi K3 (NovitaAI)",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/kimi-k3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/deepseek-v3.2": {
"id": "novita/deepseek-v3.2",
"name": "DeepSeek V3.2 (NovitaAI)",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 65536
},
"cost": {
"input": 0.269,
"output": 0.4,
"cache_read": 0.1345
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/deepseek-v3.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3.6-35b-a3b": {
"id": "novita/qwen3.6-35b-a3b",
"name": "Qwen3.6 35B A3B (NovitaAI)",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 64000
},
"cost": {
"input": 0.248,
"output": 1.485
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3.6-35b-a3b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3-max": {
"id": "novita/qwen3-max",
"name": "Qwen3 Max (NovitaAI)",
"description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.845,
"output": 3.38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-5.3-flash": {
"id": "novita/glm-5.3-flash",
"name": "GLM-5.3 Flash (NovitaAI)",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/glm-5.3-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3-vl-30b-a3b-instruct": {
"id": "novita/qwen3-vl-30b-a3b-instruct",
"name": "Qwen3 VL 30B A3B Instruct (NovitaAI)",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-10-05",
"last_updated": "2025-10-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3-vl-30b-a3b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3-vl-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/llama-4-maverick-17b-instruct": {
"id": "novita/llama-4-maverick-17b-instruct",
"name": "Llama 4 Maverick 17B Instruct (NovitaAI)",
"description": "Open multimodal Llama for strong reasoning with efficient everyday serving",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 8192
},
"cost": {
"input": 0.27,
"output": 0.85
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/llama-4-maverick-17b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/llama-4-maverick-17b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-4.5v": {
"id": "novita/glm-4.5v",
"name": "GLM-4.5V (NovitaAI)",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-08-11",
"last_updated": "2025-08-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 16000
},
"cost": {
"input": 0.6,
"output": 1.8,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/glm-4.5v\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-4.5v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3.8-flash": {
"id": "novita/qwen3.8-flash",
"name": "Qwen3.8 Flash (NovitaAI)",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.47,
"cache_read": 0.016
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3.8-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/kimi-k2": {
"id": "novita/kimi-k2",
"name": "Kimi K2 (NovitaAI)",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-11",
"last_updated": "2025-07-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.57,
"output": 2.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/kimi-k2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/kimi-k2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/gemma-4-31b-it": {
"id": "novita/gemma-4-31b-it",
"name": "Gemma 4 31B IT (NovitaAI)",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.14,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/gemma-4-31b-it\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-5": {
"id": "novita/glm-5",
"name": "GLM-5 (NovitaAI)",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202800,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/glm-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3.8-max": {
"id": "novita/qwen3.8-max",
"name": "Qwen3.8 Max (NovitaAI)",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3.8-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-5.1": {
"id": "novita/glm-5.1",
"name": "GLM-5.1 (NovitaAI)",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 1.38,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/glm-5.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3-vl-235b-a22b-thinking": {
"id": "novita/qwen3-vl-235b-a22b-thinking",
"name": "Qwen3 VL 235B A22B Thinking (NovitaAI)",
"description": "Qwen vision-language thinking model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.98,
"output": 3.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3-vl-235b-a22b-thinking\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3-vl-235b-a22b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3-vl-235b-a22b-instruct": {
"id": "novita/qwen3-vl-235b-a22b-instruct",
"name": "Qwen3 VL 235B A22B Instruct (NovitaAI)",
"description": "Qwen vision-language instruct model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3-vl-235b-a22b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3-vl-235b-a22b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3-235b-a22b-instruct-2507": {
"id": "novita/qwen3-235b-a22b-instruct-2507",
"name": "Qwen3 235B A22B Instruct 2507 (NovitaAI)",
"description": "Updated large open Qwen3 MoE instruct model for multilingual chat, coding, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-21",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.09,
"output": 0.58
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3-235b-a22b-instruct-2507\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3-235b-a22b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/qwen3-coder-480b-a35b-instruct": {
"id": "novita/qwen3-coder-480b-a35b-instruct",
"name": "Qwen3 Coder 480B A35B Instruct (NovitaAI)",
"description": "Open Qwen coding heavyweight for repository reasoning and agentic engineering",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.38,
"output": 1.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/qwen3-coder-480b-a35b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/qwen3-coder-480b-a35b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-5.3": {
"id": "novita/glm-5.3",
"name": "GLM-5.3 (NovitaAI)",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/glm-5.3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/llama-3.3-70b-instruct": {
"id": "novita/llama-3.3-70b-instruct",
"name": "Llama 3.3 70B Instruct (NovitaAI)",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 120000
},
"cost": {
"input": 0.135,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/llama-3-70b-instruct": {
"id": "novita/llama-3-70b-instruct",
"name": "Llama 3 70B Instruct (NovitaAI)",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2024-04-18",
"last_updated": "2024-04-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 8000
},
"cost": {
"input": 0.51,
"output": 0.74
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/llama-3-70b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/llama-3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/mimo-v2.5": {
"id": "novita/mimo-v2.5",
"name": "MiMo V2.5 (NovitaAI)",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.168,
"output": 0.336,
"cache_read": 0.0034,
"tiers": [
{
"input": 0.8,
"output": 4,
"cache_read": 0.16,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 0.8,
"output": 4,
"cache_read": 0.16
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/mimo-v2.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/mimo-v2.5-pro": {
"id": "novita/mimo-v2.5-pro",
"name": "MiMo V2.5 Pro (NovitaAI)",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.522,
"output": 1.044,
"cache_read": 0.0043,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/novita/mimo-v2.5-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"novita/mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ranoai/deepseek-v4-flash": {
"id": "ranoai/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (RanoAI)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 393216
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/ranoai/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"ranoai/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inference.net/llama-3.2-11b-instruct": {
"id": "inference.net/llama-3.2-11b-instruct",
"name": "Llama 3.2 11B Instruct (Inference.net)",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2024-09-25",
"last_updated": "2024-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.07,
"output": 0.33
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/inference.net/llama-3.2-11b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"inference.net/llama-3.2-11b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sakana/fugu-ultra": {
"id": "sakana/fugu-ultra",
"name": "Fugu Ultra (Sakana AI)",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/sakana/fugu-ultra\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"sakana/fugu-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sakana/fugu-max": {
"id": "sakana/fugu-max",
"name": "Fugu Max (Sakana AI)",
"description": "Multi-agent model for routing expert agents across complex analytical tasks",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-11",
"last_updated": "2026-09-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/sakana/fugu-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"sakana/fugu-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sakana/fugu-ultra-v2.0": {
"id": "sakana/fugu-ultra-v2.0",
"name": "Fugu Ultra v2.0 (Sakana AI)",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-11",
"last_updated": "2026-09-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/sakana/fugu-ultra-v2.0\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"sakana/fugu-ultra-v2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/gemma-4-26b-a4b-it": {
"id": "deepinfra/gemma-4-26b-a4b-it",
"name": "Gemma 4 26B A4B IT (DeepInfra)",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.07,
"output": 0.34
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/qwen3.5-9b": {
"id": "deepinfra/qwen3.5-9b",
"name": "Qwen3.5 9B (DeepInfra)",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/qwen3.5-9b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/qwen3.5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/ling-3.0-flash": {
"id": "deepinfra/ling-3.0-flash",
"name": "InclusionAI Ling 3.0 Flash (DeepInfra)",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "ling",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-02",
"last_updated": "2026-08-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.06,
"output": 0.18,
"cache_read": 0.012
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/ling-3.0-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/ling-3.0-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/deepseek-v4-flash": {
"id": "deepinfra/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (DeepInfra)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 393216
},
"cost": {
"input": 0.08,
"output": 0.18,
"cache_read": 0.016
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/hy3": {
"id": "deepinfra/hy3",
"name": "Hy3 (DeepInfra)",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 192000,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.58,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/hy3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/deepseek-v3.2": {
"id": "deepinfra/deepseek-v3.2",
"name": "DeepSeek V3.2 (DeepInfra)",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 160000,
"output": 65536
},
"cost": {
"input": 0.26,
"output": 0.38,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/deepseek-v3.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/nemotron-3-ultra-550b": {
"id": "deepinfra/nemotron-3-ultra-550b",
"name": "Nemotron 3 Ultra 550B (DeepInfra)",
"description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
"family": "nemotron",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.5,
"output": 2.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/nemotron-3-ultra-550b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/nemotron-3-ultra-550b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/qwen3-vl-30b-a3b-instruct": {
"id": "deepinfra/qwen3-vl-30b-a3b-instruct",
"name": "Qwen3 VL 30B A3B Instruct (DeepInfra)",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-10-05",
"last_updated": "2025-10-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/qwen3-vl-30b-a3b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/qwen3-vl-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/gemma-4-31b-it": {
"id": "deepinfra/gemma-4-31b-it",
"name": "Gemma 4 31B IT (DeepInfra)",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.13,
"output": 0.38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/gemma-4-31b-it\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/glm-5.1": {
"id": "deepinfra/glm-5.1",
"name": "GLM-5.1 (DeepInfra)",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 198000,
"output": 65536
},
"cost": {
"input": 1.05,
"output": 3.5,
"cache_read": 0.205
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/glm-5.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/qwen3-vl-235b-a22b-instruct": {
"id": "deepinfra/qwen3-vl-235b-a22b-instruct",
"name": "Qwen3 VL 235B A22B Instruct (DeepInfra)",
"description": "Qwen vision-language instruct model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.88,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/qwen3-vl-235b-a22b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/qwen3-vl-235b-a22b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/deepseek-v4-pro": {
"id": "deepinfra/deepseek-v4-pro",
"name": "DeepSeek V4 Pro (DeepInfra)",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 64000
},
"cost": {
"input": 1.3,
"output": 2.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/deepseek-v4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/mimo-v2.5": {
"id": "deepinfra/mimo-v2.5",
"name": "MiMo V2.5 (DeepInfra)",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.4,
"output": 2,
"cache_read": 0.08,
"tiers": [
{
"input": 0.8,
"output": 4,
"cache_read": 0.16,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 0.8,
"output": 4,
"cache_read": 0.16
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/mimo-v2.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepinfra/mimo-v2.5-pro": {
"id": "deepinfra/mimo-v2.5-pro",
"name": "MiMo V2.5 Pro (DeepInfra)",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": false,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 16384
},
"cost": {
"input": 1,
"output": 3,
"cache_read": 0.2,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepinfra/mimo-v2.5-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepinfra/mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure-ai-foundry/grok-4-1-fast-non-reasoning": {
"id": "azure-ai-foundry/grok-4-1-fast-non-reasoning",
"name": "Grok 4.1 Fast Non-Reasoning (Azure AI Foundry)",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 0.2,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure-ai-foundry/grok-4-1-fast-non-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure-ai-foundry/grok-4-1-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure-ai-foundry/grok-4-1-fast-reasoning": {
"id": "azure-ai-foundry/grok-4-1-fast-reasoning",
"name": "Grok 4.1 Fast Reasoning (Azure AI Foundry)",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 0.2,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure-ai-foundry/grok-4-1-fast-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure-ai-foundry/grok-4-1-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure-ai-foundry/grok-4-3": {
"id": "azure-ai-foundry/grok-4-3",
"name": "Grok 4.3 (Azure AI Foundry)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 20000,
"output": 8192
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure-ai-foundry/grok-4-3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure-ai-foundry/grok-4-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot/kimi-k2.7-code-highspeed": {
"id": "moonshot/kimi-k2.7-code-highspeed",
"name": "Kimi K2.7 Code Highspeed (Moonshot AI)",
"description": "Lower-latency Kimi Code variant for interactive edits and coding-agent loops",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.9,
"output": 8,
"cache_read": 0.38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/moonshot/kimi-k2.7-code-highspeed\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshot/kimi-k2.7-code-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot/kimi-k2.6": {
"id": "moonshot/kimi-k2.6",
"name": "Kimi K2.6 (Moonshot AI)",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/moonshot/kimi-k2.6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshot/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot/kimi-k2.7-code": {
"id": "moonshot/kimi-k2.7-code",
"name": "Kimi K2.7 Code (Moonshot AI)",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/moonshot/kimi-k2.7-code\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshot/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot/kimi-k3": {
"id": "moonshot/kimi-k3",
"name": "Kimi K3 (Moonshot AI)",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/moonshot/kimi-k3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshot/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot/kimi-k2.5": {
"id": "moonshot/kimi-k2.5",
"name": "Kimi K2.5 (Moonshot AI)",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/moonshot/kimi-k2.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshot/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"consensusprotocol/Qwen3.8-27B": {
"id": "consensusprotocol/Qwen3.8-27B",
"name": "Qwen3.8 27B (Consensus Protocol)",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/consensusprotocol/Qwen3.8-27B\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"consensusprotocol/Qwen3.8-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"consensusprotocol/deepseek-v4-flash": {
"id": "consensusprotocol/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (Consensus Protocol)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1050000,
"output": 393216
},
"cost": {
"input": 0.05,
"output": 0.1,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/consensusprotocol/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"consensusprotocol/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"consensusprotocol/gpt-oss-20b": {
"id": "consensusprotocol/gpt-oss-20b",
"name": "GPT OSS 20B (Consensus Protocol)",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 32768
},
"cost": {
"input": 0.04,
"output": 0.19,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/consensusprotocol/gpt-oss-20b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"consensusprotocol/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"consensusprotocol/glm-5.3-flash": {
"id": "consensusprotocol/glm-5.3-flash",
"name": "GLM-5.3 Flash (Consensus Protocol)",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/consensusprotocol/glm-5.3-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"consensusprotocol/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"consensusprotocol/gemma-4-31b-it": {
"id": "consensusprotocol/gemma-4-31b-it",
"name": "Gemma 4 31B IT (Consensus Protocol)",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.25,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/consensusprotocol/gemma-4-31b-it\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"consensusprotocol/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5-nano": {
"id": "azure/gpt-5-nano",
"name": "GPT-5 Nano (Azure)",
"description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
"family": "gpt-nano",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5-nano\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-4.1-nano": {
"id": "azure/gpt-4.1-nano",
"name": "GPT-4.1 Nano (Azure)",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-4.1-nano\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.1-codex-mini": {
"id": "azure/gpt-5.1-codex-mini",
"name": "GPT-5.1 Codex mini (Azure)",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.1-codex-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.1-codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.1-codex": {
"id": "azure/gpt-5.1-codex",
"name": "GPT-5.1 Codex (Azure)",
"description": "Codex GPT for repository edits, code review, and practical software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 272000
},
"cost": {
"input": 1.25,
"output": 10
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.1-codex\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.6-sol": {
"id": "azure/gpt-5.6-sol",
"name": "GPT-5.6 Sol (Azure)",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.6-sol\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.2-codex": {
"id": "azure/gpt-5.2-codex",
"name": "GPT-5.2 Codex (Azure)",
"description": "Code-specialist GPT for repository edits, reviews, and long-running software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.2-codex\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.2-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-6-astra": {
"id": "azure/gpt-6-astra",
"name": "GPT-6 Astra (Azure)",
"description": "GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.",
"family": "gpt-astra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-04-30",
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-6-astra\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-6-astra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.2-pro": {
"id": "azure/gpt-5.2-pro",
"name": "GPT-5.2 Pro (Azure)",
"description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 272000
},
"cost": {
"input": 21,
"output": 168
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.2-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-4.1-mini": {
"id": "azure/gpt-4.1-mini",
"name": "GPT-4.1 Mini (Azure)",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-4.1-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.4": {
"id": "azure/gpt-5.4",
"name": "GPT-5.4 (Azure)",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.4\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-4-turbo": {
"id": "azure/gpt-4-turbo",
"name": "GPT-4 Turbo (Azure)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 10,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-4-turbo\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-4-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.1": {
"id": "azure/gpt-5.1",
"name": "GPT-5.1 (Azure)",
"description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/o1": {
"id": "azure/o1",
"name": "o1 (Azure)",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": true,
"temperature": false,
"knowledge": "2023-09",
"release_date": "2024-12-05",
"last_updated": "2024-12-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 15,
"output": 60,
"cache_read": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/o1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/o1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-4o": {
"id": "azure/gpt-4o",
"name": "GPT-4o (Azure)",
"description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-4o\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.6-luna": {
"id": "azure/gpt-5.6-luna",
"name": "GPT-5.6 Luna (Azure)",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.6-luna\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.3-codex": {
"id": "azure/gpt-5.3-codex",
"name": "GPT-5.3 Codex (Azure)",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.3-codex\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-4.1": {
"id": "azure/gpt-4.1",
"name": "GPT-4.1 (Azure)",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-4.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.4-nano": {
"id": "azure/gpt-5.4-nano",
"name": "GPT-5.4 Nano (Azure)",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.4-nano\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.4-mini": {
"id": "azure/gpt-5.4-mini",
"name": "GPT-5.4 Mini (Azure)",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.4-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-3.5-turbo": {
"id": "azure/gpt-3.5-turbo",
"name": "GPT-3.5 Turbo (Azure)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2021-09-01",
"release_date": "2023-03-01",
"last_updated": "2023-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16385,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-3.5-turbo\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-3.5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5-mini": {
"id": "azure/gpt-5-mini",
"name": "GPT-5 Mini (Azure)",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-oss-120b": {
"id": "azure/gpt-oss-120b",
"name": "GPT OSS 120B (Azure)",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-oss-120b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.4-pro": {
"id": "azure/gpt-5.4-pro",
"name": "GPT-5.4 Pro (Azure)",
"description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.6-terra": {
"id": "azure/gpt-5.6-terra",
"name": "GPT-5.6 Terra (Azure)",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.6-terra\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-4": {
"id": "azure/gpt-4",
"name": "GPT-4 (Azure)",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-11",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 30,
"output": 60
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-4\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.2": {
"id": "azure/gpt-5.2",
"name": "GPT-5.2 (Azure)",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5": {
"id": "azure/gpt-5",
"name": "GPT-5 (Azure)",
"description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/o4-mini": {
"id": "azure/o4-mini",
"name": "o4 Mini (Azure)",
"description": "Fast o-series model for compact reasoning, coding, and tool use",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/o4-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/o4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/o3-mini": {
"id": "azure/o3-mini",
"name": "o3 Mini (Azure)",
"description": "Smaller o-series reasoner for economical coding, math, and planning tasks",
"family": "o-mini",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2024-12-20",
"last_updated": "2025-01-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/o3-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/o3": {
"id": "azure/o3",
"name": "o3 (Azure)",
"description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
"family": "o",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/o3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure/gpt-5.5": {
"id": "azure/gpt-5.5",
"name": "GPT-5.5 (Azure)",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure/gpt-5.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4.1-flash": {
"id": "deepseek/deepseek-v4.1-flash",
"name": "DeepSeek V4.1 Flash (DeepSeek)",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1050000,
"output": 393216
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepseek/deepseek-v4.1-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4.1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro": {
"id": "deepseek/deepseek-v4-pro",
"name": "DeepSeek V4 Pro (DeepSeek)",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1050000,
"output": 393216
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.003625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/deepseek/deepseek-v4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-nano": {
"id": "openai/gpt-5-nano",
"name": "GPT-5 Nano (OpenAI)",
"description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
"family": "gpt-nano",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5-nano\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-nano": {
"id": "openai/gpt-4.1-nano",
"name": "GPT-4.1 Nano (OpenAI)",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-4.1-nano\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-pro": {
"id": "openai/gpt-5-pro",
"name": "GPT-5 Pro (OpenAI)",
"description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 272000
},
"cost": {
"input": 15,
"output": 120
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini-transcribe": {
"id": "openai/gpt-4o-mini-transcribe",
"name": "GPT-4o Mini Transcribe (OpenAI)",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-03-20",
"last_updated": "2025-03-20",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16000,
"output": 2000
},
"cost": {
"input": 1.25,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-4o-mini-transcribe\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini-transcribe\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-sol": {
"id": "openai/gpt-5.6-sol",
"name": "GPT-5.6 Sol (OpenAI)",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.6-sol\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-6-astra": {
"id": "openai/gpt-6-astra",
"name": "GPT-6 Astra (OpenAI)",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-09-03",
"last_updated": "2026-09-03",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-6-astra\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-6-astra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-pro": {
"id": "openai/gpt-5.2-pro",
"name": "GPT-5.2 Pro (OpenAI)",
"description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 272000
},
"cost": {
"input": 21,
"output": 168
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.2-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-mini": {
"id": "openai/gpt-4.1-mini",
"name": "GPT-4.1 Mini (OpenAI)",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-4.1-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4": {
"id": "openai/gpt-5.4",
"name": "GPT-5.4 (OpenAI)",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.4\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-transcribe": {
"id": "openai/gpt-4o-transcribe",
"name": "GPT-4o Transcribe (OpenAI)",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-03-20",
"last_updated": "2025-03-20",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16000,
"output": 2000
},
"cost": {
"input": 2.5,
"output": 10
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-4o-transcribe\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-transcribe\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4-turbo": {
"id": "openai/gpt-4-turbo",
"name": "GPT-4 Turbo (OpenAI)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 10,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-4-turbo\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1": {
"id": "openai/gpt-5.1",
"name": "GPT-5.1 (OpenAI)",
"description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o1": {
"id": "openai/o1",
"name": "o1 (OpenAI)",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": true,
"temperature": false,
"knowledge": "2023-09",
"release_date": "2024-12-05",
"last_updated": "2024-12-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 15,
"output": 60,
"cache_read": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/o1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/o1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o": {
"id": "openai/gpt-4o",
"name": "GPT-4o (OpenAI)",
"description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-4o\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-luna": {
"id": "openai/gpt-5.6-luna",
"name": "GPT-5.6 Luna (OpenAI)",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.6-luna\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-codex": {
"id": "openai/gpt-5.3-codex",
"name": "GPT-5.3 Codex (OpenAI)",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.3-codex\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini": {
"id": "openai/gpt-4o-mini",
"name": "GPT-4o Mini (OpenAI)",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-4o-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1": {
"id": "openai/gpt-4.1",
"name": "GPT-4.1 (OpenAI)",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-4.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-nano": {
"id": "openai/gpt-5.4-nano",
"name": "GPT-5.4 Nano (OpenAI)",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.4-nano\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5-pro": {
"id": "openai/gpt-5.5-pro",
"name": "GPT-5.5 Pro (OpenAI)",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180,
"tiers": [
{
"input": 60,
"output": 270,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 60,
"output": 270
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.5-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-mini": {
"id": "openai/gpt-5.4-mini",
"name": "GPT-5.4 Mini (OpenAI)",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.4-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-3.5-turbo": {
"id": "openai/gpt-3.5-turbo",
"name": "GPT-3.5 Turbo (OpenAI)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2021-09-01",
"release_date": "2023-03-01",
"last_updated": "2023-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16385,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-3.5-turbo\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-3.5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-mini": {
"id": "openai/gpt-5-mini",
"name": "GPT-5 Mini (OpenAI)",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-pro": {
"id": "openai/gpt-5.4-pro",
"name": "GPT-5.4 Pro (OpenAI)",
"description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-terra": {
"id": "openai/gpt-5.6-terra",
"name": "GPT-5.6 Terra (OpenAI)",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.6-terra\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4": {
"id": "openai/gpt-4",
"name": "GPT-4 (OpenAI)",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-11",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 30,
"output": 60
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-4\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2": {
"id": "openai/gpt-5.2",
"name": "GPT-5.2 (OpenAI)",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5": {
"id": "openai/gpt-5",
"name": "GPT-5 (OpenAI)",
"description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o4-mini": {
"id": "openai/o4-mini",
"name": "o4 Mini (OpenAI)",
"description": "Fast o-series model for compact reasoning, coding, and tool use",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/o4-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/o4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-mini": {
"id": "openai/o3-mini",
"name": "o3 Mini (OpenAI)",
"description": "Smaller o-series reasoner for economical coding, math, and planning tasks",
"family": "o-mini",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2024-12-20",
"last_updated": "2025-01-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/o3-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3": {
"id": "openai/o3",
"name": "o3 (OpenAI)",
"description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
"family": "o",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/o3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT-5.5 (OpenAI)",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/openai/gpt-5.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-contributor/muse-spark-1.2-contributor": {
"id": "meta-contributor/muse-spark-1.2-contributor",
"name": "Muse Spark 1.2 Contributor (Meta Contributor)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-06",
"last_updated": "2026-08-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.002
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/meta-contributor/muse-spark-1.2-contributor\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta-contributor/muse-spark-1.2-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-contributor/muse-spark-1.3-contributor": {
"id": "meta-contributor/muse-spark-1.3-contributor",
"name": "Muse Spark 1.3 Contributor (Meta Contributor)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.002
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/meta-contributor/muse-spark-1.3-contributor\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta-contributor/muse-spark-1.3-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4": {
"id": "xai/grok-4",
"name": "Grok 4 (xAI)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-09",
"last_updated": "2025-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/xai/grok-4\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4-5": {
"id": "xai/grok-4-5",
"name": "Grok 4.5 (xAI)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/xai/grok-4-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-build-0-1": {
"id": "xai/grok-build-0-1",
"name": "Grok Build 0.1 (xAI)",
"description": "Grok coding model for agentic engineering, edits, and codebase workflows",
"family": "grok-build",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-05-20",
"last_updated": "2026-05-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 1,
"output": 2,
"cache_read": 0.2,
"tiers": [
{
"input": 2,
"output": 4,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2,
"output": 4,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/xai/grok-build-0-1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-build-0-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4-3": {
"id": "xai/grok-4-3",
"name": "Grok 4.3 (xAI)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/xai/grok-4-3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4-20-beta-0309-reasoning": {
"id": "xai/grok-4-20-beta-0309-reasoning",
"name": "Grok 4.20 Beta Reasoning (0309) (xAI)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/xai/grok-4-20-beta-0309-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4-20-beta-0309-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4-6": {
"id": "xai/grok-4-6",
"name": "Grok 4.6 (xAI)",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/xai/grok-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4-20-beta-0309-non-reasoning": {
"id": "xai/grok-4-20-beta-0309-non-reasoning",
"name": "Grok 4.20 Beta Non-Reasoning (0309) (xAI)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/xai/grok-4-20-beta-0309-non-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4-20-beta-0309-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.7": {
"id": "zai/glm-4.7",
"name": "GLM-4.7 (Z AI)",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-4.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.5-air": {
"id": "zai/glm-4.5-air",
"name": "GLM-4.5 Air (Z AI)",
"description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
"family": "glm-air",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 98304
},
"cost": {
"input": 0.2,
"output": 1.1,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-4.5-air\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.5-air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.6": {
"id": "zai/glm-4.6",
"name": "GLM-4.6 (Z AI)",
"description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-4.6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.6v": {
"id": "zai/glm-4.6v",
"name": "GLM-4.6V (Z AI)",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16000
},
"cost": {
"input": 0.3,
"output": 0.9,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-4.6v\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.6v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.6v-flashx": {
"id": "zai/glm-4.6v-flashx",
"name": "GLM-4.6V FlashX (Z AI)",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16000
},
"cost": {
"input": 0.04,
"output": 0.4,
"cache_read": 0.004
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-4.6v-flashx\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.6v-flashx\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.2": {
"id": "zai/glm-5.2",
"name": "GLM-5.2 (Z AI)",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.5-x": {
"id": "zai/glm-4.5-x",
"name": "GLM-4.5 X (Z AI)",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 2.2,
"output": 8.9,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-4.5-x\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.5-x\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.5-airx": {
"id": "zai/glm-4.5-airx",
"name": "GLM-4.5 AirX (Z AI)",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 1.1,
"output": 4.5,
"cache_read": 0.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-4.5-airx\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.5-airx\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.3-flash": {
"id": "zai/glm-5.3-flash",
"name": "GLM-5.3 Flash (Z AI)",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-5.3-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.5": {
"id": "zai/glm-4.5",
"name": "GLM-4.5 (Z AI)",
"description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 98304
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-4.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.5v": {
"id": "zai/glm-4.5v",
"name": "GLM-4.5V (Z AI)",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-08-11",
"last_updated": "2025-08-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16000
},
"cost": {
"input": 0.6,
"output": 1.8,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-4.5v\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.5v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.7-flashx": {
"id": "zai/glm-4.7-flashx",
"name": "GLM-4.7 FlashX (Z AI)",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.07,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-4.7-flashx\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.7-flashx\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5": {
"id": "zai/glm-5",
"name": "GLM-5 (Z AI)",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202800,
"output": 131100
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4-32b-0414-128k": {
"id": "zai/glm-4-32b-0414-128k",
"name": "GLM-4 32B (0414-128k) (Z AI)",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.1,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-4-32b-0414-128k\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4-32b-0414-128k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.1": {
"id": "zai/glm-5.1",
"name": "GLM-5.1 (Z AI)",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-5.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.3": {
"id": "zai/glm-5.3",
"name": "GLM-5.3 (Z AI)",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/zai/glm-5.3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure-anthropic/claude-opus-5": {
"id": "azure-anthropic/claude-opus-5",
"name": "Claude Opus 5 (Azure Anthropic)",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure-anthropic/claude-opus-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure-anthropic/claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure-anthropic/claude-opus-4-6": {
"id": "azure-anthropic/claude-opus-4-6",
"name": "Claude Opus 4.6 (Azure Anthropic)",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure-anthropic/claude-opus-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure-anthropic/claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure-anthropic/claude-opus-4-7": {
"id": "azure-anthropic/claude-opus-4-7",
"name": "Claude Opus 4.7 (Azure Anthropic)",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure-anthropic/claude-opus-4-7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure-anthropic/claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure-anthropic/claude-fable-5": {
"id": "azure-anthropic/claude-fable-5",
"name": "Claude Fable 5 (Azure Anthropic)",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure-anthropic/claude-fable-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure-anthropic/claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure-anthropic/claude-opus-4-8": {
"id": "azure-anthropic/claude-opus-4-8",
"name": "Claude Opus 4.8 (Azure Anthropic)",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure-anthropic/claude-opus-4-8\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure-anthropic/claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"azure-anthropic/claude-sonnet-5": {
"id": "azure-anthropic/claude-sonnet-5",
"name": "Claude Sonnet 5 (Azure Anthropic)",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/azure-anthropic/claude-sonnet-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"azure-anthropic/claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fireworks/deepseek-v4-flash": {
"id": "fireworks/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (Fireworks AI)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393216
},
"cost": {
"input": 0.22,
"output": 0.66,
"cache_read": 0.007
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/fireworks/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fireworks/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fireworks/kimi-k3": {
"id": "fireworks/kimi-k3",
"name": "Kimi K3 (Fireworks AI)",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1040384,
"output": 1040384
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/fireworks/kimi-k3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fireworks/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fireworks/kimi-k3-fast": {
"id": "fireworks/kimi-k3-fast",
"name": "Kimi K3 Fast (Fireworks AI)",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1040384,
"output": 1040384
},
"cost": {
"input": 4.5,
"output": 22.5,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/fireworks/kimi-k3-fast\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fireworks/kimi-k3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fireworks/deepseek-v4-pro": {
"id": "fireworks/deepseek-v4-pro",
"name": "DeepSeek V4 Pro (Fireworks AI)",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393216
},
"cost": {
"input": 1.32,
"output": 3.96,
"cache_read": 0.044
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/fireworks/deepseek-v4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fireworks/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/ministral-14b-2512": {
"id": "mistral/ministral-14b-2512",
"name": "Ministral 14B (Mistral AI)",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.2,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/mistral/ministral-14b-2512\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/ministral-14b-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/codestral-2508": {
"id": "mistral/codestral-2508",
"name": "Codestral (Mistral AI)",
"description": "Mistral coding model for code completion, generation, and developer workflows",
"family": "codestral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-30",
"last_updated": "2025-07-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.3,
"output": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/mistral/codestral-2508\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/codestral-2508\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-small-2506": {
"id": "mistral/mistral-small-2506",
"name": "Mistral Small 3.2 (Mistral AI)",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-06-20",
"last_updated": "2025-06-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.1,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/mistral/mistral-small-2506\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-small-2506\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/devstral-2512": {
"id": "mistral/devstral-2512",
"name": "Devstral 2 (Mistral AI)",
"description": "Mistral's coding-agent model for repository work, terminal tasks, and software fixes",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2025-12-09",
"last_updated": "2025-12-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.4,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/mistral/devstral-2512\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/devstral-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-large-2512": {
"id": "mistral/mistral-large-2512",
"name": "Mistral Large 3 (Mistral AI)",
"description": "Mistral's largest general model for enterprise agents, coding, and multilingual reasoning",
"family": "mistral-large",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-11",
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.5,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/mistral/mistral-large-2512\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-large-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/ministral-3b-2512": {
"id": "mistral/ministral-3b-2512",
"name": "Ministral 3B (Mistral AI)",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/mistral/ministral-3b-2512\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/ministral-3b-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-large-latest": {
"id": "mistral/mistral-large-latest",
"name": "Mistral Large Latest (Mistral AI)",
"description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
"family": "mistral-large",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-11",
"release_date": "2024-11-01",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 262144
},
"cost": {
"input": 4,
"output": 12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/mistral/mistral-large-latest\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-large-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/ministral-8b-2512": {
"id": "mistral/ministral-8b-2512",
"name": "Ministral 8B (Mistral AI)",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.15,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/mistral/ministral-8b-2512\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/ministral-8b-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras/glm-4.7": {
"id": "cerebras/glm-4.7",
"name": "GLM-4.7 (Cerebras)",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 2.25,
"output": 2.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/cerebras/glm-4.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"cerebras/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras/gemma-4-31b-it": {
"id": "cerebras/gemma-4-31b-it",
"name": "Gemma 4 31B IT (Cerebras)",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.99,
"output": 1.49
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/cerebras/gemma-4-31b-it\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"cerebras/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras/qwen3-235b-a22b-instruct-2507": {
"id": "cerebras/qwen3-235b-a22b-instruct-2507",
"name": "Qwen3 235B A22B Instruct 2507 (Cerebras)",
"description": "Updated large open Qwen3 MoE instruct model for multilingual chat, coding, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-21",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 8192
},
"cost": {
"input": 0.6,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/cerebras/qwen3-235b-a22b-instruct-2507\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"cerebras/qwen3-235b-a22b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras/gpt-oss-120b": {
"id": "cerebras/gpt-oss-120b",
"name": "GPT OSS 120B (Cerebras)",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.35,
"output": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/cerebras/gpt-oss-120b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"cerebras/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras/llama-3.3-70b-instruct": {
"id": "cerebras/llama-3.3-70b-instruct",
"name": "Llama 3.3 70B Instruct (Cerebras)",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.85,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/cerebras/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"cerebras/llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/sonar": {
"id": "perplexity/sonar",
"name": "Sonar (Perplexity)",
"description": "Fast web-grounded Sonar for current answers, citations, and lightweight retrieval",
"family": "sonar",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 130000,
"output": 4096
},
"cost": {
"input": 1,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/perplexity/sonar\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/sonar\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/sonar-reasoning-pro": {
"id": "perplexity/sonar-reasoning-pro",
"name": "Sonar Reasoning Pro (Perplexity)",
"description": "Web-grounded Sonar for multi-step research questions that need cited reasoning",
"family": "sonar-reasoning",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 2,
"output": 8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/perplexity/sonar-reasoning-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/sonar-reasoning-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/sonar-pro": {
"id": "perplexity/sonar-pro",
"name": "Sonar Pro (Perplexity)",
"description": "Deeper Sonar search model with broader retrieval and stronger synthesis",
"family": "sonar-pro",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 3,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/perplexity/sonar-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/sonar-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"runware/kimi-k2.6": {
"id": "runware/kimi-k2.6",
"name": "Kimi K2.6 (Runware)",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 3.05,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/runware/kimi-k2.6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"runware/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"runware/glm-5.2": {
"id": "runware/glm-5.2",
"name": "GLM-5.2 (Runware)",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1024000,
"output": 128000
},
"cost": {
"input": 0.8,
"output": 2.55,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/runware/glm-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"runware/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"runware/deepseek-v4-flash": {
"id": "runware/deepseek-v4-flash",
"name": "DeepSeek V4 Flash (Runware)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.076,
"output": 0.153,
"cache_read": 0.014
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/runware/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"runware/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"runware/kimi-k3": {
"id": "runware/kimi-k3",
"name": "Kimi K3 (Runware)",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/runware/kimi-k3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"runware/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"runware/glm-5.3-flash": {
"id": "runware/glm-5.3-flash",
"name": "GLM-5.3 Flash (Runware)",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/runware/glm-5.3-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"runware/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"runware/gemma-4-31b-it": {
"id": "runware/gemma-4-31b-it",
"name": "Gemma 4 31B IT (Runware)",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.102,
"output": 0.297,
"cache_read": 0.012
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/runware/gemma-4-31b-it\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"runware/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"runware/deepseek-v4-pro": {
"id": "runware/deepseek-v4-pro",
"name": "DeepSeek V4 Pro (Runware)",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.961,
"output": 1.922,
"cache_read": 0.079
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/runware/deepseek-v4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"runware/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"runware/gpt-oss-120b": {
"id": "runware/gpt-oss-120b",
"name": "GPT OSS 120B (Runware)",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.032,
"output": 0.14,
"cache_read": 0.032
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/runware/gpt-oss-120b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"runware/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"runware/glm-5.3": {
"id": "runware/glm-5.3",
"name": "GLM-5.3 (Runware)",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway-providers/runware/glm-5.3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"runware/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"llama": {
"id": "llama",
"name": "Llama",
"baseURL": "https://api.llama.com/compat/v1/",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"LLAMA_API_KEY"
],
"doc": "https://llama.developer.meta.com/docs/models",
"modelCount": 7,
"models": {
"cerebras-llama-4-scout-17b-16e-instruct": {
"id": "cerebras-llama-4-scout-17b-16e-instruct",
"name": "Cerebras-Llama-4-Scout-17B-16E-Instruct",
"description": "Open multimodal Llama model for long-context analysis and efficient agents",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llama/cerebras-llama-4-scout-17b-16e-instruct\", apiKey: processEnvironment[\"LLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llama.com/compat/v1/\")!,\n apiKey: processEnvironment[\"LLAMA_API_KEY\"]\n)\nlet session = provider.model(\"cerebras-llama-4-scout-17b-16e-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-4-maverick-17b-128e-instruct-fp8": {
"id": "llama-4-maverick-17b-128e-instruct-fp8",
"name": "Llama-4-Maverick-17B-128E-Instruct-FP8",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llama/llama-4-maverick-17b-128e-instruct-fp8\", apiKey: processEnvironment[\"LLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llama.com/compat/v1/\")!,\n apiKey: processEnvironment[\"LLAMA_API_KEY\"]\n)\nlet session = provider.model(\"llama-4-maverick-17b-128e-instruct-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"groq-llama-4-maverick-17b-128e-instruct": {
"id": "groq-llama-4-maverick-17b-128e-instruct",
"name": "Groq-Llama-4-Maverick-17B-128E-Instruct",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llama/groq-llama-4-maverick-17b-128e-instruct\", apiKey: processEnvironment[\"LLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llama.com/compat/v1/\")!,\n apiKey: processEnvironment[\"LLAMA_API_KEY\"]\n)\nlet session = provider.model(\"groq-llama-4-maverick-17b-128e-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-4-scout-17b-16e-instruct-fp8": {
"id": "llama-4-scout-17b-16e-instruct-fp8",
"name": "Llama-4-Scout-17B-16E-Instruct-FP8",
"description": "Open multimodal Llama model for long-context analysis and efficient agents",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llama/llama-4-scout-17b-16e-instruct-fp8\", apiKey: processEnvironment[\"LLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llama.com/compat/v1/\")!,\n apiKey: processEnvironment[\"LLAMA_API_KEY\"]\n)\nlet session = provider.model(\"llama-4-scout-17b-16e-instruct-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras-llama-4-maverick-17b-128e-instruct": {
"id": "cerebras-llama-4-maverick-17b-128e-instruct",
"name": "Cerebras-Llama-4-Maverick-17B-128E-Instruct",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llama/cerebras-llama-4-maverick-17b-128e-instruct\", apiKey: processEnvironment[\"LLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llama.com/compat/v1/\")!,\n apiKey: processEnvironment[\"LLAMA_API_KEY\"]\n)\nlet session = provider.model(\"cerebras-llama-4-maverick-17b-128e-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-70b-instruct": {
"id": "llama-3.3-70b-instruct",
"name": "Llama-3.3-70B-Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llama/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"LLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llama.com/compat/v1/\")!,\n apiKey: processEnvironment[\"LLAMA_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-8b-instruct": {
"id": "llama-3.3-8b-instruct",
"name": "Llama-3.3-8B-Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llama/llama-3.3-8b-instruct\", apiKey: processEnvironment[\"LLAMA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llama.com/compat/v1/\")!,\n apiKey: processEnvironment[\"LLAMA_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.3-8b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"alibaba-token-plan": {
"id": "alibaba-token-plan",
"name": "Alibaba Token Plan",
"baseURL": "https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ALIBABA_TOKEN_PLAN_API_KEY"
],
"doc": "https://www.alibabacloud.com/help/en/model-studio/token-plan-overview",
"modelCount": 26,
"models": {
"qwen3.7-max": {
"id": "qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/qwen3.7-max\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"happyhorse-1.1-r2v": {
"id": "happyhorse-1.1-r2v",
"name": "HappyHorse 1.1 Reference-to-Video",
"description": "Video model for reference-guided video generation",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-07-17",
"last_updated": "2026-07-17",
"modalities": {
"input": [
"image",
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/happyhorse-1.1-r2v\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"happyhorse-1.1-r2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro-0813": {
"id": "deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max-preview": {
"id": "qwen3.8-max-preview",
"name": "Qwen3.8 Max Preview",
"description": "Preview Qwen flagship for million-token multimodal reasoning and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens",
"min": 0,
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-07-19",
"last_updated": "2026-07-19",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"status": "deprecated",
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/qwen3.8-max-preview\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-plus": {
"id": "qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 131072
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/qwen3.6-plus\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"wan2.7-image-pro": {
"id": "wan2.7-image-pro",
"name": "Wan2.7 Image Pro",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/wan2.7-image-pro\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"wan2.7-image-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/kimi-k2.6\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"happyhorse-1.1-t2v": {
"id": "happyhorse-1.1-t2v",
"name": "HappyHorse 1.1 Text-to-Video",
"description": "Video model for prompt-driven text-to-video generation",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-07-17",
"last_updated": "2026-07-17",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/happyhorse-1.1-t2v\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"happyhorse-1.1-t2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/glm-5.2\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/deepseek-v4-flash\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code": {
"id": "kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/kimi-k2.7-code\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-image-2.0-pro": {
"id": "qwen-image-2.0-pro",
"name": "Qwen Image 2.0 Pro",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/qwen-image-2.0-pro\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen-image-2.0-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3.2": {
"id": "deepseek-v3.2",
"name": "DeepSeek V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-03",
"last_updated": "2025-12-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/deepseek-v3.2\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.5": {
"id": "MiniMax-M2.5",
"name": "MiniMax-M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"input": 196601,
"output": 32768
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/MiniMax-M2.5\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"happyhorse-1.1-i2v": {
"id": "happyhorse-1.1-i2v",
"name": "HappyHorse 1.1 Image-to-Video",
"description": "Video model for image-to-video generation",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-07-17",
"last_updated": "2026-07-17",
"modalities": {
"input": [
"image",
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/happyhorse-1.1-i2v\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"happyhorse-1.1-i2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-image-2.0": {
"id": "qwen-image-2.0",
"name": "Qwen Image 2.0",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/qwen-image-2.0\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen-image-2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-flash": {
"id": "qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 131072
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/qwen3.6-flash\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-flash": {
"id": "qwen3.8-flash",
"name": "Qwen3.8 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens",
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/qwen3.8-flash\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5": {
"id": "glm-5",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 202752,
"output": 16384
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/glm-5\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max": {
"id": "qwen3.8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens",
"min": 0,
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/qwen3.8-max\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 98304
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/kimi-k2.5\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 128000
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/glm-5.1\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-plus": {
"id": "qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/qwen3.7-plus\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/deepseek-v4-pro\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"wan2.7-image": {
"id": "wan2.7-image",
"name": "Wan2.7 Image",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-token-plan/wan2.7-image\", apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_TOKEN_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"wan2.7-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"neuralwatt": {
"id": "neuralwatt",
"name": "Neuralwatt",
"baseURL": "https://api.neuralwatt.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"NEURALWATT_API_KEY"
],
"doc": "https://portal.neuralwatt.com/docs",
"modelCount": 21,
"models": {
"glm-5.2-short-fast-flex": {
"id": "glm-5.2-short-fast-flex",
"name": "GLM 5.2 Short Fast Flex",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-17",
"last_updated": "2026-06-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 199984,
"output": 32000
},
"cost": {
"input": 0.9425,
"output": 2.925,
"cache_read": 0.09425
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/glm-5.2-short-fast-flex\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-short-fast-flex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-short-flex": {
"id": "glm-5.2-short-flex",
"name": "GLM 5.2 Short Flex",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-17",
"last_updated": "2026-06-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 199984,
"output": 32000
},
"cost": {
"input": 0.9425,
"output": 2.925,
"cache_read": 0.09425
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/glm-5.2-short-flex\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-short-flex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-flex": {
"id": "glm-5.2-flex",
"name": "GLM 5.2 Flex",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-17",
"last_updated": "2026-06-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048560,
"output": 1048560
},
"cost": {
"input": 0.9425,
"output": 2.925,
"cache_read": 0.09425
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/glm-5.2-flex\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-flex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM 5.2",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-17",
"last_updated": "2026-06-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048560,
"output": 1048560
},
"cost": {
"input": 1.45,
"output": 4.5,
"cache_read": 0.145
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/glm-5.2\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048560,
"output": 65536
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/deepseek-v4-flash\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code": {
"id": "kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262128,
"output": 262128
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.095
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/kimi-k2.7-code\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code-flex": {
"id": "kimi-k2.7-code-flex",
"name": "Kimi K2.7 Code Flex",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262128,
"output": 262128
},
"cost": {
"input": 0.6175,
"output": 2.6,
"cache_read": 0.06175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/kimi-k2.7-code-flex\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code-flex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code-fast": {
"id": "kimi-k2.7-code-fast",
"name": "Kimi K2.7 Code Fast",
"description": "Kimi K2.7 Code with reasoning capped to a short budget for lower latency; reasoning cannot be disabled on this model",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262128,
"output": 262128
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.095
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/kimi-k2.7-code-fast\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-short-fast": {
"id": "glm-5.2-short-fast",
"name": "GLM 5.2 Short Fast",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-17",
"last_updated": "2026-06-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 199984,
"output": 32000
},
"cost": {
"input": 1.45,
"output": 4.5,
"cache_read": 0.145
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/glm-5.2-short-fast\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-short-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048560,
"output": 1048560
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/kimi-k3\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3-fast": {
"id": "kimi-k3-fast",
"name": "Kimi K3 Fast",
"description": "Kimi K3 with thinking disabled for low-latency tool calling, vision, and JSON work",
"family": "kimi-k3",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048560,
"output": 1048560
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/kimi-k3-fast\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-fast": {
"id": "glm-5.2-fast",
"name": "GLM 5.2 Fast",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-17",
"last_updated": "2026-06-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048560,
"output": 1048560
},
"cost": {
"input": 1.45,
"output": 4.5,
"cache_read": 0.145
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/glm-5.2-fast\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3-flex": {
"id": "kimi-k3-flex",
"name": "Kimi K3 Flex",
"description": "Kimi K3 on the flex tier: discounted, best-effort latency, requests may be held under load",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048560,
"output": 1048560
},
"cost": {
"input": 1.95,
"output": 9.75,
"cache_read": 0.195
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/kimi-k3-flex\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3-flex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-35b-fast": {
"id": "qwen3.6-35b-fast",
"name": "Qwen3.6 35B Fast",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "qwen3.6",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-04-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131056,
"output": 131056
},
"cost": {
"input": 0.29,
"output": 1.15,
"cache_read": 0.029
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/qwen3.6-35b-fast\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-35b-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b": {
"id": "gemma-4-31b",
"name": "Gemma 4 31B",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262128,
"output": 16384
},
"cost": {
"input": 0.144,
"output": 0.42,
"cache_read": 0.0144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/gemma-4-31b\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048560,
"output": 393216
},
"status": "beta",
"cost": {
"input": 1,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/deepseek-v4-pro\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-flex": {
"id": "deepseek-v4-flash-flex",
"name": "DeepSeek V4 Flash Flex",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048560,
"output": 65536
},
"cost": {
"input": 0.091,
"output": 0.182,
"cache_read": 0.0182
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/deepseek-v4-flash-flex\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-flex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM 5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048560,
"output": 1048560
},
"status": "beta",
"cost": {
"input": 1.45,
"output": 4.5,
"cache_read": 0.145
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/glm-5.3\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-35b": {
"id": "qwen3.6-35b",
"name": "Qwen3.6 35B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131056,
"output": 131056
},
"cost": {
"input": 0.29,
"output": 1.15,
"cache_read": 0.029
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/qwen3.6-35b\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-35b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-3.8-27b": {
"id": "qwen-3.8-27b",
"name": "Qwen3.8 27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262128,
"output": 65536
},
"status": "beta",
"cost": {
"input": 0.45,
"output": 3.2,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/qwen-3.8-27b\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"qwen-3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-short": {
"id": "glm-5.2-short",
"name": "GLM 5.2 Short",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-17",
"last_updated": "2026-06-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 199984,
"output": 32000
},
"cost": {
"input": 1.45,
"output": 4.5,
"cache_read": 0.145
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"neuralwatt/glm-5.2-short\", apiKey: processEnvironment[\"NEURALWATT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.neuralwatt.com/v1\")!,\n apiKey: processEnvironment[\"NEURALWATT_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-short\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"abliteration-ai": {
"id": "abliteration-ai",
"name": "abliteration.ai",
"baseURL": "https://api.abliteration.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ABLIT_KEY"
],
"doc": "https://docs.abliteration.ai/models",
"modelCount": 3,
"models": {
"abliterated-model-large": {
"id": "abliterated-model-large",
"name": "Abliterated Model Large",
"description": "GLM-5.2 model abliterated and finetuned for cyber, ML red teaming, and agent testing",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
},
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-25",
"last_updated": "2026-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 999990
},
"cost": {
"input": 5,
"output": 5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"abliteration-ai/abliterated-model-large\", apiKey: processEnvironment[\"ABLIT_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.abliteration.ai/v1\")!,\n apiKey: processEnvironment[\"ABLIT_KEY\"]\n)\nlet session = provider.model(\"abliterated-model-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"abliterated-model-large-v2": {
"id": "abliterated-model-large-v2",
"name": "Abliterated Model Large V2",
"description": "GLM-5.3 model abliterated and finetuned for cyber, ML red teaming, and agent testing",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-29",
"last_updated": "2026-08-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 999990
},
"cost": {
"input": 5,
"output": 5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"abliteration-ai/abliterated-model-large-v2\", apiKey: processEnvironment[\"ABLIT_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.abliteration.ai/v1\")!,\n apiKey: processEnvironment[\"ABLIT_KEY\"]\n)\nlet session = provider.model(\"abliterated-model-large-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"abliterated-model": {
"id": "abliterated-model",
"name": "Abliterated Model",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
},
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-01-06",
"last_updated": "2026-07-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 150000,
"input": 150000,
"output": 8192
},
"cost": {
"input": 3,
"output": 3,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"abliteration-ai/abliterated-model\", apiKey: processEnvironment[\"ABLIT_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.abliteration.ai/v1\")!,\n apiKey: processEnvironment[\"ABLIT_KEY\"]\n)\nlet session = provider.model(\"abliterated-model\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"clarifai": {
"id": "clarifai",
"name": "Clarifai",
"baseURL": "https://api.clarifai.com/v2/ext/openai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"CLARIFAI_PAT"
],
"doc": "https://docs.clarifai.com/compute/inference/",
"modelCount": 12,
"models": {
"qwen/qwenCoder/models/Qwen3-Coder-30B-A3B-Instruct": {
"id": "qwen/qwenCoder/models/Qwen3-Coder-30B-A3B-Instruct",
"name": "Qwen3 Coder 30B A3B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-31",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.11458,
"output": 0.74812
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/qwen/qwenCoder/models/Qwen3-Coder-30B-A3B-Instruct\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"qwen/qwenCoder/models/Qwen3-Coder-30B-A3B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwenLM/models/Qwen3-30B-A3B-Instruct-2507": {
"id": "qwen/qwenLM/models/Qwen3-30B-A3B-Instruct-2507",
"name": "Qwen3 30B A3B Instruct 2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-30",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.3,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/qwen/qwenLM/models/Qwen3-30B-A3B-Instruct-2507\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"qwen/qwenLM/models/Qwen3-30B-A3B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwenLM/models/Qwen3-30B-A3B-Thinking-2507": {
"id": "qwen/qwenLM/models/Qwen3-30B-A3B-Thinking-2507",
"name": "Qwen3 30B A3B Thinking 2507",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-31",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.36,
"output": 1.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/qwen/qwenLM/models/Qwen3-30B-A3B-Thinking-2507\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"qwen/qwenLM/models/Qwen3-30B-A3B-Thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"clarifai/main/models/mm-poly-8b": {
"id": "clarifai/main/models/mm-poly-8b",
"name": "MM Poly 8B",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"family": "mm-poly",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-06",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 4096
},
"cost": {
"input": 0.658,
"output": 1.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/clarifai/main/models/mm-poly-8b\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"clarifai/main/models/mm-poly-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/deepseek-ocr/models/DeepSeek-OCR": {
"id": "deepseek-ai/deepseek-ocr/models/DeepSeek-OCR",
"name": "DeepSeek OCR",
"description": "OCR model for extracting structured text from documents and screenshots",
"family": "deepseek",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-10-20",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0.2,
"output": 0.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/deepseek-ai/deepseek-ocr/models/DeepSeek-OCR\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"deepseek-ai/deepseek-ocr/models/DeepSeek-OCR\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/completion/models/Ministral-3-14B-Reasoning-2512": {
"id": "mistralai/completion/models/Ministral-3-14B-Reasoning-2512",
"name": "Ministral 3 14B Reasoning 2512",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2025-12-01",
"last_updated": "2025-12-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 2.5,
"output": 1.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/mistralai/completion/models/Ministral-3-14B-Reasoning-2512\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"mistralai/completion/models/Ministral-3-14B-Reasoning-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/completion/models/Ministral-3-3B-Reasoning-2512": {
"id": "mistralai/completion/models/Ministral-3-3B-Reasoning-2512",
"name": "Ministral 3 3B Reasoning 2512",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-12",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.039,
"output": 0.54825
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/mistralai/completion/models/Ministral-3-3B-Reasoning-2512\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"mistralai/completion/models/Ministral-3-3B-Reasoning-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimaxai/chat-completion/models/MiniMax-M2_5-high-throughput": {
"id": "minimaxai/chat-completion/models/MiniMax-M2_5-high-throughput",
"name": "MiniMax-M2.5 High Throughput",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/minimaxai/chat-completion/models/MiniMax-M2_5-high-throughput\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"minimaxai/chat-completion/models/MiniMax-M2_5-high-throughput\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/chat-completion/models/gpt-oss-120b-high-throughput": {
"id": "openai/chat-completion/models/gpt-oss-120b-high-throughput",
"name": "GPT OSS 120B High Throughput",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.09,
"output": 0.36
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/openai/chat-completion/models/gpt-oss-120b-high-throughput\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"openai/chat-completion/models/gpt-oss-120b-high-throughput\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/chat-completion/models/gpt-oss-20b": {
"id": "openai/chat-completion/models/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-12-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.045,
"output": 0.18
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/openai/chat-completion/models/gpt-oss-20b\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"openai/chat-completion/models/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/chat-completion/models/Kimi-K2_6": {
"id": "moonshotai/chat-completion/models/Kimi-K2_6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/moonshotai/chat-completion/models/Kimi-K2_6\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"moonshotai/chat-completion/models/Kimi-K2_6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"arcee_ai/AFM/models/trinity-mini": {
"id": "arcee_ai/AFM/models/trinity-mini",
"name": "Trinity Mini",
"description": "Reasoning-tuned 26B MoE model with 3B active parameters for agents, tools, and multi-step workloads",
"family": "trinity",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-12-01",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.045,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"clarifai/arcee_ai/AFM/models/trinity-mini\", apiKey: processEnvironment[\"CLARIFAI_PAT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.clarifai.com/v2/ext/openai/v1\")!,\n apiKey: processEnvironment[\"CLARIFAI_PAT\"]\n)\nlet session = provider.model(\"arcee_ai/AFM/models/trinity-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"morph": {
"id": "morph",
"name": "Morph",
"baseURL": "https://api.morphllm.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"MORPH_API_KEY"
],
"doc": "https://docs.morphllm.com/api-reference/introduction",
"modelCount": 3,
"models": {
"morph-v3-large": {
"id": "morph-v3-large",
"name": "Morph v3 Large",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "morph",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-08-15",
"last_updated": "2024-08-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 32000
},
"cost": {
"input": 0.9,
"output": 1.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"morph/morph-v3-large\", apiKey: processEnvironment[\"MORPH_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.morphllm.com/v1\")!,\n apiKey: processEnvironment[\"MORPH_API_KEY\"]\n)\nlet session = provider.model(\"morph-v3-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"morph-v3-fast": {
"id": "morph-v3-fast",
"name": "Morph v3 Fast",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "morph",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-08-15",
"last_updated": "2024-08-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16000,
"output": 16000
},
"cost": {
"input": 0.8,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"morph/morph-v3-fast\", apiKey: processEnvironment[\"MORPH_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.morphllm.com/v1\")!,\n apiKey: processEnvironment[\"MORPH_API_KEY\"]\n)\nlet session = provider.model(\"morph-v3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"auto": {
"id": "auto",
"name": "Auto",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"family": "auto",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-06-01",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 32000
},
"cost": {
"input": 0.85,
"output": 1.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"morph/auto\", apiKey: processEnvironment[\"MORPH_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.morphllm.com/v1\")!,\n apiKey: processEnvironment[\"MORPH_API_KEY\"]\n)\nlet session = provider.model(\"auto\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"aihubmix": {
"id": "aihubmix",
"name": "AIHubMix",
"baseURL": "",
"npm": "@aihubmix/ai-sdk-provider",
"swiftDriver": "openaiChat",
"env": [
"AIHUBMIX_API_KEY"
],
"doc": "https://docs.aihubmix.com",
"modelCount": 78,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75,
"tiers": [
{
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/claude-sonnet-4-6\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-max": {
"id": "qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"output": 64000
},
"cost": {
"input": 1.69,
"output": 5.07,
"cache_read": 0.169,
"cache_write": 2.1125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/qwen3.7-max\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code-highspeed": {
"id": "kimi-k2.7-code-highspeed",
"name": "Kimi K2.7 Code Highspeed",
"description": "Lower-latency Kimi Code variant for interactive edits and coding-agent loops",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 1.9,
"output": 7.999,
"cache_read": 0.32167
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/kimi-k2.7-code-highspeed\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro-preview-customtools": {
"id": "gemini-3.1-pro-preview-customtools",
"name": "Gemini 3.1 Pro Preview Custom Tools",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gemini-3.1-pro-preview-customtools\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-pro-preview-customtools\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro-0813": {
"id": "deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.6918,
"output": 2.0754,
"cache_read": 0.023058
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-0-lite-260428": {
"id": "doubao-seed-2-0-lite-260428",
"name": "Doubao Seed 2.0 Lite 260428",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-28",
"last_updated": "2026-04-28",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.08,
"output": 0.51,
"cache_read": 0.01692,
"input_audio": 1.269,
"tiers": [
{
"input": 0.13,
"output": 0.76,
"cache_read": 0.02536,
"input_audio": 1.902,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.25,
"output": 1.52,
"cache_read": 0.05072,
"input_audio": 3.804,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/doubao-seed-2-0-lite-260428\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-lite-260428\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.142,
"output": 0.284,
"cache_read": 0.0284
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"coding-minimax-m2.7": {
"id": "coding-minimax-m2.7",
"name": "Coding MiniMax M2.7",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 128100
},
"cost": {
"input": 0.2,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/coding-minimax-m2.7\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"coding-minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"coding-glm-5.1": {
"id": "coding-glm-5.1",
"name": "Coding GLM 5.1",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-11",
"last_updated": "2026-04-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.06,
"output": 0.22,
"cache_read": 0.013
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/coding-glm-5.1\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"coding-glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7-think": {
"id": "claude-opus-4-7-think",
"name": "Claude Opus 4.7 Thinking",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/claude-opus-4-7-think\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7-think\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex-mini": {
"id": "gpt-5.1-codex-mini",
"name": "GPT-5.1 Codex mini",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.1-codex-mini\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.3": {
"id": "grok-4.3",
"name": "Grok 4.3",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-01",
"last_updated": "2026-05-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/grok-4.3\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-plus": {
"id": "qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-05-09",
"last_updated": "2026-05-09",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"output": 64000
},
"cost": {
"input": 0.28,
"output": 1.69,
"cache_read": 0.0282,
"cache_write": 0.3525,
"tiers": [
{
"input": 1.13,
"output": 6.77,
"cache_read": 0.1128,
"cache_write": 1.41,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 1.13,
"output": 6.77,
"cache_read": 0.1128,
"cache_write": 1.41
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/qwen3.6-plus\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex": {
"id": "gpt-5.1-codex",
"name": "GPT-5.1 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.1-codex\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-0-mini-260428": {
"id": "doubao-seed-2-0-mini-260428",
"name": "Doubao Seed 2.0 Mini 260428",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-28",
"last_updated": "2026-04-28",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.03,
"output": 0.28,
"cache_read": 0.00564,
"input_audio": 0.423,
"tiers": [
{
"input": 0.06,
"output": 0.56,
"cache_read": 0.01128,
"input_audio": 0.846,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.11,
"output": 1.13,
"cache_read": 0.02256,
"input_audio": 1.692,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/doubao-seed-2-0-mini-260428\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-mini-260428\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-sol": {
"id": "gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.6-sol\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5": {
"id": "claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/claude-opus-5\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.7": {
"id": "minimax-m2.7",
"name": "MiniMax M2.7",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 128000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/minimax-m2.7\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-0-code-preview": {
"id": "doubao-seed-2-0-code-preview",
"name": "Doubao Seed 2.0 Code Preview",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.48,
"output": 2.41,
"cache_read": 0.09644,
"tiers": [
{
"input": 0.72,
"output": 3.62,
"cache_read": 0.144656,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 1.45,
"output": 7.23,
"cache_read": 0.28932,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/doubao-seed-2-0-code-preview\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-code-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi-mimo-v2.5-free": {
"id": "xiaomi-mimo-v2.5-free",
"name": "Xiaomi MiMo-V2.5 (free)",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo-v2.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-05-13",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/xiaomi-mimo-v2.5-free\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi-mimo-v2.5-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/kimi-k2.6\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro-preview": {
"id": "gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"coding-xiaomi-mimo-v2.5-pro": {
"id": "coding-xiaomi-mimo-v2.5-pro",
"name": "Coding Xiaomi MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo-v2.5-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-05-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.2,
"output": 0.6,
"cache_read": 0.04,
"tiers": [
{
"input": 0.4,
"output": 1.2,
"cache_read": 0.08,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 0.4,
"output": 1.2,
"cache_read": 0.08
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/coding-xiaomi-mimo-v2.5-pro\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"coding-xiaomi-mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2-codex": {
"id": "gpt-5.2-codex",
"name": "GPT-5.2 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.2-codex\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.2-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.1268,
"output": 3.9438,
"cache_read": 0.2817
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/glm-5.2\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code": {
"id": "kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.95,
"output": 3.9995,
"cache_read": 0.160835
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/kimi-k2.7-code\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.5": {
"id": "grok-4.5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/grok-4.5\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alicloud-deepseek-v4-pro": {
"id": "alicloud-deepseek-v4-pro",
"name": "DeepSeek V4 Pro (Alibaba Cloud)",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 1.69,
"output": 3.38,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/alicloud-deepseek-v4-pro\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"alicloud-deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"provider": {
"body": {
"service_tier": "priority"
}
}
}
}
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.4\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite": {
"id": "gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"cache_write": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-build-0.1": {
"id": "grok-build-0.1",
"name": "Grok Build 0.1",
"description": "Fast Grok coding model tuned for agentic engineering and iterative edits",
"family": "grok-build",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 1,
"output": 2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/grok-build-0.1\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"grok-build-0.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi-mimo-v2.5-pro-free": {
"id": "xiaomi-mimo-v2.5-pro-free",
"name": "Xiaomi MiMo-V2.5-Pro (free)",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo-v2.5-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-05-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/xiaomi-mimo-v2.5-pro-free\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi-mimo-v2.5-pro-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4.1-flash": {
"id": "deepseek-v4.1-flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.155,
"output": 0.62,
"cache_read": 0.0031
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/deepseek-v4.1-flash\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4.1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1": {
"id": "gpt-5.1",
"name": "GPT-5.1",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.1\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deep-deepseek-v4-pro": {
"id": "deep-deepseek-v4-pro",
"name": "DeepSeek V4 Pro (DeepSeek)",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.478,
"output": 0.956,
"cache_read": 0.004302
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/deep-deepseek-v4-pro\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"deep-deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deep-deepseek-v4-flash": {
"id": "deep-deepseek-v4-flash",
"name": "DeepSeek V4 Flash (DeepSeek)",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.154,
"output": 0.308,
"cache_read": 0.0308
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/deep-deepseek-v4-flash\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"deep-deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/claude-opus-4-6\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash": {
"id": "gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gemini-3.5-flash\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi-mimo-v2.5": {
"id": "xiaomi-mimo-v2.5",
"name": "Xiaomi MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo-v2.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-05-13",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.44,
"output": 2.2,
"cache_read": 0.088,
"tiers": [
{
"input": 0.88,
"output": 4.4,
"cache_read": 0.176,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 0.88,
"output": 4.4,
"cache_read": 0.176
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/xiaomi-mimo-v2.5\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi-mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-luna": {
"id": "gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 1,
"output": 6,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.6-luna\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-flash": {
"id": "qwen3.7-flash",
"name": "Qwen3.7 Flash",
"description": "Lightweight multimodal Qwen model for high-throughput text, image, and video tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"input": 991000,
"output": 64000
},
"cost": {
"input": 0.0282,
"output": 0.1128,
"cache_read": 0.00564,
"cache_write": 0.03525
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/qwen3.7-flash\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7": {
"id": "claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/claude-opus-4-7\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alicloud-deepseek-v4-flash": {
"id": "alicloud-deepseek-v4-flash",
"name": "DeepSeek V4 Flash (Alibaba Cloud)",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/alicloud-deepseek-v4-flash\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"alicloud-deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/kimi-k3\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.3-codex": {
"id": "gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.3-codex\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-glm-5.1": {
"id": "zai-glm-5.1",
"name": "GLM-5.1 (Z.ai)",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-27",
"last_updated": "2026-03-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.845,
"output": 3.38,
"cache_read": 0.183112
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/zai-glm-5.1\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"zai-glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.11268,
"output": 0.39438,
"cache_read": 0.02817
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/glm-5.3-flash\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 11,
"output": 55,
"cache_read": 1.1,
"cache_write": 13.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/claude-fable-5\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-2.4t-a95b": {
"id": "qwen3.8-2.4t-a95b",
"name": "Qwen3.8 2.4T A95B",
"description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/qwen3.8-2.4t-a95b\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-2.4t-a95b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8-think": {
"id": "claude-opus-4-8-think",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/claude-opus-4-8-think\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8-think\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"coding-xiaomi-mimo-v2.5": {
"id": "coding-xiaomi-mimo-v2.5",
"name": "Coding Xiaomi MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo-v2.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-05-13",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.08,
"output": 0.4,
"cache_read": 0.016,
"tiers": [
{
"input": 0.16,
"output": 0.8,
"cache_read": 0.032,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 0.16,
"output": 0.8,
"cache_read": 0.032
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/coding-xiaomi-mimo-v2.5\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"coding-xiaomi-mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-flash": {
"id": "qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"output": 64000
},
"cost": {
"input": 0.17,
"output": 1.01,
"cache_read": 0.0169,
"cache_write": 0.21125,
"tiers": [
{
"input": 0.68,
"output": 4.06,
"cache_read": 0.0676,
"cache_write": 0.845,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 0.68,
"output": 4.06,
"cache_read": 0.0676,
"cache_write": 0.845
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/qwen3.6-flash\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-mini": {
"id": "gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15
},
"provider": {
"body": {
"service_tier": "priority"
}
}
}
}
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.4-mini\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"coding-minimax-m2.7-free": {
"id": "coding-minimax-m2.7-free",
"name": "Coding MiniMax M2.7 (Free)",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax-free",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 128100
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/coding-minimax-m2.7-free\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"coding-minimax-m2.7-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-0-pro": {
"id": "doubao-seed-2-0-pro",
"name": "Doubao Seed 2.0 Pro",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.48,
"output": 2.41,
"cache_read": 0.09644,
"tiers": [
{
"input": 0.72,
"output": 3.62,
"cache_read": 0.144656,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 1.45,
"output": 7.23,
"cache_read": 0.28932,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/doubao-seed-2-0-pro\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-max-preview": {
"id": "qwen3.6-max-preview",
"name": "Qwen3.6 Max Preview",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "qwen3.6",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-05-09",
"last_updated": "2026-05-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 240000,
"output": 64000
},
"cost": {
"input": 1.27,
"output": 7.61,
"cache_read": 0.1268,
"cache_write": 1.585,
"tiers": [
{
"input": 2.11,
"output": 12.67,
"cache_read": 0.2112,
"cache_write": 2.64,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/qwen3.6-max-preview\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.6": {
"id": "grok-4.6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/grok-4.6\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"grok-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alicloud-glm-5.1": {
"id": "alicloud-glm-5.1",
"name": "GLM-5.1 (Alibaba Cloud)",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-27",
"last_updated": "2026-03-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.84,
"output": 3.38,
"cache_read": 0.169,
"cache_write": 1.05625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/alicloud-glm-5.1\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"alicloud-glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max": {
"id": "qwen3.8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"output": 128000
},
"cost": {
"input": 1.69,
"output": 5.07,
"cache_read": 0.169,
"cache_write": 2.1125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/qwen3.8-max\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "Kimi K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/kimi-k2.5\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-flash-preview": {
"id": "gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"tiers": [
{
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gemini-3-flash-preview\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-6-think": {
"id": "claude-sonnet-4-6-think",
"name": "Claude Sonnet 4.6 Thinking",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75,
"tiers": [
{
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/claude-sonnet-4-6-think\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6-think\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-plus": {
"id": "qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"output": 64000
},
"cost": {
"input": 0.282,
"output": 1.128,
"cache_read": 0.0564,
"cache_write": 0.3525
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/qwen3.7-plus\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/claude-opus-4-8\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.7-flash": {
"id": "gemini-3.7-flash",
"name": "Gemini 3.7 Flash",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gemini-3.7-flash\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"hy3-preview": {
"id": "hy3-preview",
"name": "Hy3 Preview",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-20",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.17,
"output": 0.566661,
"cache_read": 0.051
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/hy3-preview\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"hy3-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-03-20",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gemini-2.5-pro\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-terra": {
"id": "gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.6-terra\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.1268,
"output": 3.9438,
"cache_read": 0.2817
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/glm-5.3\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5v-turbo": {
"id": "glm-5v-turbo",
"name": "GLM 5 Vision Turbo",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glmv",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-05-09",
"last_updated": "2026-05-09",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.7042,
"output": 3.09848,
"cache_read": 0.169008
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/glm-5v-turbo\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"glm-5v-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2": {
"id": "gpt-5.2",
"name": "GPT-5.2",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.2\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-03-20",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"input_audio": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gemini-2.5-flash\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi-mimo-v2.5-pro": {
"id": "xiaomi-mimo-v2.5-pro",
"name": "Xiaomi MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo-v2.5-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-05-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.1,
"output": 3.3,
"cache_read": 0.22,
"tiers": [
{
"input": 2.2,
"output": 6.6,
"cache_read": 0.44,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2.2,
"output": 6.6,
"cache_read": 0.44
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/xiaomi-mimo-v2.5-pro\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi-mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5": {
"id": "claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/claude-sonnet-5\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6-think": {
"id": "claude-opus-4-6-think",
"name": "Claude Opus 4.6 Thinking",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/claude-opus-4-6-think\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6-think\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"coding-glm-5.1-free": {
"id": "coding-glm-5.1-free",
"name": "Coding GLM 5.1 (free)",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm-free",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-11",
"last_updated": "2026-04-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/coding-glm-5.1-free\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"coding-glm-5.1-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"coding-minimax-m2.7-highspeed": {
"id": "coding-minimax-m2.7-highspeed",
"name": "Coding MiniMax M2.7 Highspeed",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 128100
},
"cost": {
"input": 0.2,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/coding-minimax-m2.7-highspeed\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"coding-minimax-m2.7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "GPT-5.5",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 12.5,
"output": 75,
"cache_read": 1.25
},
"provider": {
"body": {
"service_tier": "priority"
}
}
}
}
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aihubmix/gpt-5.5\", apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AIHUBMIX_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"chutes": {
"id": "chutes",
"name": "Chutes",
"baseURL": "https://llm.chutes.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"CHUTES_API_KEY"
],
"doc": "https://llm.chutes.ai/v1/models",
"modelCount": 14,
"models": {
"Nemotron-3-Nano-Omni-30B-TEE": {
"id": "Nemotron-3-Nano-Omni-30B-TEE",
"name": "Nemotron 3 Nano Omni 30B TEE",
"description": "Omni-modal model for text, vision, audio, and multimodal agent tasks",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-07-23",
"last_updated": "2026-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 0
},
"cost": {
"input": 0.0245,
"output": 0.0978,
"cache_read": 0.0024499999999999995
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/Nemotron-3-Nano-Omni-30B-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"Nemotron-3-Nano-Omni-30B-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Flash-0731-TEE": {
"id": "deepseek-ai/DeepSeek-V4-Flash-0731-TEE",
"name": "DeepSeek V4 Flash 0731 TEE",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-08-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.44,
"output": 1.32,
"cache_read": 0.04399999999999999
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/deepseek-ai/DeepSeek-V4-Flash-0731-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash-0731-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.2-TEE": {
"id": "deepseek-ai/DeepSeek-V3.2-TEE",
"name": "DeepSeek V3.2 TEE",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12",
"last_updated": "2026-06-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 1,
"output": 1,
"cache_read": 0.09999999999999998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/deepseek-ai/DeepSeek-V3.2-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.2-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31B-turbo-TEE": {
"id": "google/gemma-4-31B-turbo-TEE",
"name": "gemma 4 31B turbo TEE",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0.12,
"output": 0.37,
"cache_read": 0.011999999999999997
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/google/gemma-4-31B-turbo-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31B-turbo-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.1-TEE": {
"id": "zai-org/GLM-5.1-TEE",
"name": "GLM 5.1 TEE",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 65535
},
"cost": {
"input": 0.98,
"output": 3.08,
"cache_read": 0.09799999999999998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/zai-org/GLM-5.1-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.1-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2-TEE": {
"id": "zai-org/GLM-5.2-TEE",
"name": "GLM 5.2 TEE",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 1.25,
"output": 3.95,
"cache_read": 0.12499999999999997
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/zai-org/GLM-5.2-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.8-27B-TEE": {
"id": "Qwen/Qwen3.8-27B-TEE",
"name": "Qwen3.8 27B TEE",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-16",
"last_updated": "2026-08-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.32,
"output": 2.5,
"cache_read": 0.031999999999999994
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/Qwen/Qwen3.8-27B-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.8-27B-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-27B-TEE": {
"id": "Qwen/Qwen3.6-27B-TEE",
"name": "Qwen3.6 27B TEE",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2,
"cache_read": 0.029999999999999992
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/Qwen/Qwen3.6-27B-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-27B-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-397B-A17B-TEE": {
"id": "Qwen/Qwen3.5-397B-A17B-TEE",
"name": "Qwen3.5 397B A17B TEE",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.45,
"output": 3,
"cache_read": 0.04499999999999999
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/Qwen/Qwen3.5-397B-A17B-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-397B-A17B-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B-Thinking-2507-TEE": {
"id": "Qwen/Qwen3-235B-A22B-Thinking-2507-TEE",
"name": "Qwen3 235B A22B Thinking 2507 TEE",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07",
"last_updated": "2026-06-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.2989,
"output": 1.1957,
"cache_read": 0.029889999999999993
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/Qwen/Qwen3-235B-A22B-Thinking-2507-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B-Thinking-2507-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-32B-TEE": {
"id": "Qwen/Qwen3-32B-TEE",
"name": "Qwen3 32B TEE",
"description": "Dense open Qwen model for self-hosted chat, reasoning, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 40960
},
"cost": {
"input": 0.104,
"output": 0.416,
"cache_read": 0.010399999999999998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/Qwen/Qwen3-32B-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-32B-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"unsloth/Mistral-Nemo-Instruct-2407-TEE": {
"id": "unsloth/Mistral-Nemo-Instruct-2407-TEE",
"name": "Mistral Nemo Instruct 2407 TEE",
"description": "Efficient Mistral-NVIDIA open model for multilingual chat and local deployment",
"family": "mistral-nemo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-07-01",
"last_updated": "2024-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.0245,
"output": 0.0978,
"cache_read": 0.0024499999999999995
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/unsloth/Mistral-Nemo-Instruct-2407-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"unsloth/Mistral-Nemo-Instruct-2407-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K3-TEE": {
"id": "moonshotai/Kimi-K3-TEE",
"name": "Kimi K3 TEE",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.29999999999999993
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/moonshotai/Kimi-K3-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K3-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6-TEE": {
"id": "moonshotai/Kimi-K2.6-TEE",
"name": "Kimi K2.6 TEE",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65535
},
"cost": {
"input": 0.58,
"output": 3.4,
"cache_read": 0.05799999999999998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"chutes/moonshotai/Kimi-K2.6-TEE\", apiKey: processEnvironment[\"CHUTES_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llm.chutes.ai/v1\")!,\n apiKey: processEnvironment[\"CHUTES_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6-TEE\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"groq": {
"id": "groq",
"name": "Groq",
"baseURL": "",
"npm": "@ai-sdk/groq",
"swiftDriver": "openaiChat",
"env": [
"GROQ_API_KEY"
],
"doc": "https://console.groq.com/docs/models",
"modelCount": 16,
"models": {
"whisper-large-v3": {
"id": "whisper-large-v3",
"name": "Whisper",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "whisper",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2023-09-01",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/whisper-large-v3\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"whisper-large-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"whisper-large-v3-turbo": {
"id": "whisper-large-v3-turbo",
"name": "Whisper Large V3 Turbo",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "whisper",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-10-01",
"last_updated": "2024-10-01",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/whisper-large-v3-turbo\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"whisper-large-v3-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.1-8b-instant": {
"id": "llama-3.1-8b-instant",
"name": "Llama 3.1 8B",
"description": "Compact Llama instruction model for fast chat and local deployment",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.08
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/llama-3.1-8b-instant\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.1-8b-instant\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"allam-2-7b": {
"id": "allam-2-7b",
"name": "ALLaM-2-7b",
"description": "ALLaM-2-7b instruction tuned model by SDAIA",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-01-23",
"last_updated": "2025-01-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 4096,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/allam-2-7b\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"allam-2-7b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-70b-versatile": {
"id": "llama-3.3-70b-versatile",
"name": "Llama 3.3 70B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.59,
"output": 0.79
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/llama-3.3-70b-versatile\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.3-70b-versatile\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-27b": {
"id": "qwen/qwen3.8-27b",
"name": "Qwen3.8 27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"default",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131042,
"output": 16384
},
"cost": {
"input": 0.8,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/qwen/qwen3.8-27b\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-27b": {
"id": "qwen/qwen3.6-27b",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"default"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/qwen/qwen3.6-27b\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"groq/compound": {
"id": "groq/compound",
"name": "Compound",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"family": "groq",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-09-04",
"last_updated": "2025-09-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/groq/compound\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"groq/compound\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"groq/compound-mini": {
"id": "groq/compound-mini",
"name": "Compound Mini",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "groq",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-09-04",
"last_updated": "2025-09-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/groq/compound-mini\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"groq/compound-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-prompt-guard-2-86m": {
"id": "meta-llama/llama-prompt-guard-2-86m",
"name": "Prompt Guard 2 86M",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-05-29",
"last_updated": "2025-05-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512,
"output": 512
},
"status": "beta",
"cost": {
"input": 0.04,
"output": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/meta-llama/llama-prompt-guard-2-86m\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-prompt-guard-2-86m\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-prompt-guard-2-22m": {
"id": "meta-llama/llama-prompt-guard-2-22m",
"name": "Llama Prompt Guard 2 22M",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-05-29",
"last_updated": "2025-05-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512,
"output": 512
},
"status": "beta",
"cost": {
"input": 0.03,
"output": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/meta-llama/llama-prompt-guard-2-22m\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-prompt-guard-2-22m\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0.075,
"output": 0.3,
"cache_read": 0.0375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/openai/gpt-oss-20b\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-safeguard-20b": {
"id": "openai/gpt-oss-safeguard-20b",
"name": "Safety GPT OSS 20B",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-29",
"last_updated": "2026-06-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"status": "beta",
"cost": {
"input": 0.075,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/openai/gpt-oss-safeguard-20b\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-safeguard-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-10-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/openai/gpt-oss-120b\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"canopylabs/orpheus-v1-english": {
"id": "canopylabs/orpheus-v1-english",
"name": "Canopy Labs Orpheus V1 English",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "canopylabs",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-12-19",
"last_updated": "2025-12-19",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 4000,
"output": 50000
},
"status": "beta",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/canopylabs/orpheus-v1-english\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"canopylabs/orpheus-v1-english\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"canopylabs/orpheus-arabic-saudi": {
"id": "canopylabs/orpheus-arabic-saudi",
"name": "Canopy Labs Orpheus Arabic Saudi",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "canopylabs",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-12-16",
"last_updated": "2025-12-16",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 4000,
"output": 50000
},
"status": "beta",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"groq/canopylabs/orpheus-arabic-saudi\", apiKey: processEnvironment[\"GROQ_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GROQ_API_KEY\"]\n)\nlet session = provider.model(\"canopylabs/orpheus-arabic-saudi\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"zai-coding-plan": {
"id": "zai-coding-plan",
"name": "Z.AI Coding Plan",
"baseURL": "https://api.z.ai/api/coding/paas/v4",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ZHIPU_API_KEY"
],
"doc": "https://docs.z.ai/devpack/overview",
"modelCount": 7,
"models": {
"glm-5.2-highspeed": {
"id": "glm-5.2-highspeed",
"name": "GLM-5.2 Highspeed",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zai-coding-plan/glm-5.2-highspeed\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.z.ai/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.7": {
"id": "glm-4.7",
"name": "GLM-4.7",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zai-coding-plan/glm-4.7\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.z.ai/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zai-coding-plan/glm-5.2\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.z.ai/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-highspeed": {
"id": "glm-5.3-highspeed",
"name": "GLM-5.3 Highspeed",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zai-coding-plan/glm-5.3-highspeed\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.z.ai/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zai-coding-plan/glm-5.3-flash\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.z.ai/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5-turbo": {
"id": "glm-5-turbo",
"name": "GLM-5-Turbo",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zai-coding-plan/glm-5-turbo\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.z.ai/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zai-coding-plan/glm-5.3\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.z.ai/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"volcengine": {
"id": "volcengine",
"name": "Volcengine Ark",
"baseURL": "https://ark.cn-beijing.volces.com/api/v3",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ARK_API_KEY"
],
"doc": "https://www.volcengine.com/docs/82379/1330310",
"modelCount": 15,
"models": {
"doubao-seed-2-0-lite-260428": {
"id": "doubao-seed-2-0-lite-260428",
"name": "Seed 2.0 Lite",
"description": "Cost-efficient ByteDance Seed 2.0 model for production chat, analysis, and structured generation",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 131072
},
"cost": {
"input": 0.08906,
"output": 0.53436,
"cache_read": 0.01781,
"tiers": [
{
"input": 0.13359,
"output": 0.80154,
"cache_read": 0.02672,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.26718,
"output": 1.60308,
"cache_read": 0.05344,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-2-0-lite-260428\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-lite-260428\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-character-260628": {
"id": "doubao-seed-character-260628",
"name": "Seed Character",
"description": "ByteDance Seed model optimized for character-driven dialogue and consistent conversational behavior",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-23",
"last_updated": "2026-06-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.11875,
"output": 0.29687,
"cache_read": 0.02375,
"tiers": [
{
"input": 0.17812,
"output": 0.8906,
"cache_read": 0.02375,
"tier": {
"type": "context",
"size": 32000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-character-260628\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-character-260628\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-0-mini-260428": {
"id": "doubao-seed-2-0-mini-260428",
"name": "Seed 2.0 Mini",
"description": "Lightweight ByteDance Seed 2.0 model for low-latency multimodal reasoning and high-volume tasks",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 131072
},
"cost": {
"input": 0.02969,
"output": 0.29687,
"cache_read": 0.00594,
"tiers": [
{
"input": 0.05937,
"output": 0.59374,
"cache_read": 0.01187,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.11875,
"output": 1.18747,
"cache_read": 0.02375,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-2-0-mini-260428\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-mini-260428\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-1-pro-260628": {
"id": "doubao-seed-2-1-pro-260628",
"name": "Seed 2.1 Pro",
"description": "Flagship ByteDance Seed 2.1 model for complex multimodal reasoning, coding, and agents",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-23",
"last_updated": "2026-06-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.8906,
"output": 4.45301,
"cache_read": 0.17812
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-2-1-pro-260628\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-1-pro-260628\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-0-code-preview-260215": {
"id": "doubao-seed-2-0-code-preview-260215",
"name": "Seed 2.0 Code",
"description": "ByteDance Seed coding model for multimodal software engineering and long-running agents",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.47499,
"output": 2.37494,
"cache_read": 0.095,
"tiers": [
{
"input": 0.71248,
"output": 3.56241,
"cache_read": 0.1425,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 1.42496,
"output": 7.12482,
"cache_read": 0.28499,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-2-0-code-preview-260215\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-code-preview-260215\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-1-8-251228": {
"id": "doubao-seed-1-8-251228",
"name": "Seed 1.8",
"description": "ByteDance Seed model for multimodal reasoning, long-context analysis, and agent workflows",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-12-28",
"last_updated": "2025-12-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0.11875,
"output": 1.18747,
"cache_read": 0.02375,
"tiers": [
{
"input": 0.17812,
"output": 2.37494,
"cache_read": 0.02375,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.35624,
"output": 3.56241,
"cache_read": 0.02375,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-1-8-251228\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-1-8-251228\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-1-6-flash-250828": {
"id": "doubao-seed-1-6-flash-250828",
"name": "Seed 1.6 Flash",
"description": "Low-latency ByteDance Seed model for high-throughput chat, extraction, and lightweight tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-08-28",
"last_updated": "2025-08-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0.02227,
"output": 0.22265,
"cache_read": 0.00445,
"tiers": [
{
"input": 0.04453,
"output": 0.4453,
"cache_read": 0.00445,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.08906,
"output": 0.8906,
"cache_read": 0.00445,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-1-6-flash-250828\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-1-6-flash-250828\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-1-6-251015": {
"id": "doubao-seed-1-6-251015",
"name": "Seed 1.6",
"description": "ByteDance Seed model for long-context reasoning, instruction following, and tool-assisted tasks",
"family": "seed",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0.11875,
"output": 1.18747,
"cache_read": 0.02375,
"tiers": [
{
"input": 0.17812,
"output": 2.37494,
"cache_read": 0.02375,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.35624,
"output": 3.56241,
"cache_read": 0.02375,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-1-6-251015\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-1-6-251015\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro-ga-260813": {
"id": "deepseek-v4-pro-ga-260813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 1.3359,
"output": 4.00771,
"cache_read": 0.04453
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/deepseek-v4-pro-ga-260813\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-ga-260813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-evolving": {
"id": "doubao-seed-evolving",
"name": "Seed Evolving",
"description": "Rolling ByteDance Seed model for rapidly updated reasoning, coding, and agent capabilities",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-23",
"last_updated": "2026-06-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.8906,
"output": 4.45301,
"cache_read": 0.17812
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-evolving\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-evolving\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-0-pro-260215": {
"id": "doubao-seed-2-0-pro-260215",
"name": "Seed 2.0 Pro",
"description": "Flagship ByteDance Seed 2.0 model for complex multimodal reasoning and long-horizon agent workflows",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.47499,
"output": 2.37494,
"cache_read": 0.095,
"tiers": [
{
"input": 0.71248,
"output": 3.56241,
"cache_read": 0.1425,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 1.42496,
"output": 7.12482,
"cache_read": 0.28499,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-2-0-pro-260215\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-0-pro-260215\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-1-6-vision-250815": {
"id": "doubao-seed-1-6-vision-250815",
"name": "Seed 1.6 Vision",
"description": "ByteDance Seed multimodal model for image understanding, visual reasoning, and tool-assisted tasks",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-08-15",
"last_updated": "2025-08-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0.11875,
"output": 1.18747,
"cache_read": 0.02375,
"tiers": [
{
"input": 0.17812,
"output": 2.37494,
"cache_read": 0.02375,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.35624,
"output": 3.56241,
"cache_read": 0.02375,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-1-6-vision-250815\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-1-6-vision-250815\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5-2-260617": {
"id": "glm-5-2-260617",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.18747,
"output": 4.15615,
"cache_read": 0.29687
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/glm-5-2-260617\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"glm-5-2-260617\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"doubao-seed-2-1-turbo-260628": {
"id": "doubao-seed-2-1-turbo-260628",
"name": "Seed 2.1 Turbo",
"description": "Faster ByteDance Seed 2.1 model for multimodal reasoning and latency-sensitive agent workflows",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-23",
"last_updated": "2026-06-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.4453,
"output": 2.22651,
"cache_read": 0.08906
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/doubao-seed-2-1-turbo-260628\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"doubao-seed-2-1-turbo-260628\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-ga-260731": {
"id": "deepseek-v4-flash-ga-260731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.4453,
"output": 1.3359,
"cache_read": 0.01484
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"volcengine/deepseek-v4-flash-ga-260731\", apiKey: processEnvironment[\"ARK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ark.cn-beijing.volces.com/api/v3\")!,\n apiKey: processEnvironment[\"ARK_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-ga-260731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"sensenova": {
"id": "sensenova",
"name": "SenseNova (China)",
"baseURL": "https://token.sensenova.cn/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"SENSENOVA_API_KEY"
],
"doc": "https://platform.sensenova.cn/docs",
"modelCount": 5,
"models": {
"sensenova-6.8-flash-lite": {
"id": "sensenova-6.8-flash-lite",
"name": "SenseNova 6.8 Flash Lite",
"description": "SenseNova lightweight multimodal agent model for real-world complex tasks, data analysis, and complex information presentation",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-11",
"last_updated": "2026-08-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sensenova/sensenova-6.8-flash-lite\", apiKey: processEnvironment[\"SENSENOVA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token.sensenova.cn/v1\")!,\n apiKey: processEnvironment[\"SENSENOVA_API_KEY\"]\n)\nlet session = provider.model(\"sensenova-6.8-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sensenova/glm-5.2\", apiKey: processEnvironment[\"SENSENOVA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token.sensenova.cn/v1\")!,\n apiKey: processEnvironment[\"SENSENOVA_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sensenova/deepseek-v4-flash\", apiKey: processEnvironment[\"SENSENOVA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token.sensenova.cn/v1\")!,\n apiKey: processEnvironment[\"SENSENOVA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sensenova/kimi-k3\", apiKey: processEnvironment[\"SENSENOVA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token.sensenova.cn/v1\")!,\n apiKey: processEnvironment[\"SENSENOVA_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sensenova/deepseek-v4-pro\", apiKey: processEnvironment[\"SENSENOVA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token.sensenova.cn/v1\")!,\n apiKey: processEnvironment[\"SENSENOVA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"orcarouter": {
"id": "orcarouter",
"name": "OrcaRouter",
"baseURL": "https://api.orcarouter.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ORCAROUTER_API_KEY"
],
"doc": "https://docs.orcarouter.ai",
"modelCount": 117,
"models": {
"qwen/qwen3.7-max": {
"id": "qwen/qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 3.75,
"cache_read": 0.25,
"cache_write": 1.563
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.7-max\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-plus": {
"id": "qwen/qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"cache_write": 0.625,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.6-plus\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-27b": {
"id": "qwen/qwen3.5-27b",
"name": "Qwen3.5 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.086,
"output": 0.688
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.5-27b\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-27b": {
"id": "qwen/qwen3.8-27b",
"name": "Qwen3.8 27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.33,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.8-27b\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-35b-a3b": {
"id": "qwen/qwen3.5-35b-a3b",
"name": "Qwen3.5 35B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.057,
"output": 0.459
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.5-35b-a3b\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-flash": {
"id": "qwen/qwen3.5-flash",
"name": "Qwen3.5 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1,
"max": 81920
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.5-flash\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-397b-a17b": {
"id": "qwen/qwen3.5-397b-a17b",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.172,
"output": 1.032
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.7-flash": {
"id": "qwen/qwen3.7-flash",
"name": "Qwen3.7 Flash",
"description": "Lightweight multimodal Qwen model for high-throughput text, image, and video tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 991000,
"output": 65536
},
"cost": {
"input": 0.03,
"output": 0.13,
"cache_read": 0.006,
"cache_write": 0.038
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.7-flash\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-35b-a3b": {
"id": "qwen/qwen3.6-35b-a3b",
"name": "Qwen3.6 35B-A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.248,
"output": 1.485
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.6-35b-a3b\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-max": {
"id": "qwen/qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.359,
"output": 1.434
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3-max\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-122b-a10b": {
"id": "qwen/qwen3.5-122b-a10b",
"name": "Qwen3.5 122B-A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.115,
"output": 0.917
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.5-122b-a10b\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-122b-a10b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-flash": {
"id": "qwen/qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.6-flash\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-max": {
"id": "qwen/qwen3.8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens",
"min": 0,
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.8-max\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-235b-a22b-thinking": {
"id": "qwen/qwen3-vl-235b-a22b-thinking",
"name": "Qwen3 VL 235B A22B Thinking",
"description": "Qwen vision-language thinking model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 81920
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 40960
},
"cost": {
"input": 0.4,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3-vl-235b-a22b-thinking\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-235b-a22b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-235b-a22b-instruct": {
"id": "qwen/qwen3-vl-235b-a22b-instruct",
"name": "Qwen3 VL 235B A22B Instruct",
"description": "Qwen vision-language instruct model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3-vl-235b-a22b-instruct\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-235b-a22b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.7-plus": {
"id": "qwen/qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.35,
"output": 1.42,
"cache_read": 0.071
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.7-plus\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-plus": {
"id": "qwen/qwen3.5-plus",
"name": "Qwen3.5 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.115,
"output": 0.688,
"reasoning": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/qwen/qwen3.5-plus\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"orcarouter/free": {
"id": "orcarouter/free",
"name": "OrcaRouter Free",
"description": "Built-in router over the free tier that scores each request's difficulty and sends light work to the smaller free model and harder work to the stronger one. Priced at zero and never falls back to a paid model.",
"family": "auto",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-27",
"last_updated": "2026-08-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/orcarouter/free\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"orcarouter/free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"orcarouter/fusion-mini": {
"id": "orcarouter/fusion-mini",
"name": "OrcaRouter Fusion Mini",
"description": "Leaner two-model Fusion panel that runs Claude Opus 4.8 and GPT-5.5 in parallel on hard requests, then has a Claude Opus 4.8 judge return the strongest single answer verbatim. Easy requests fall through to a cheaper default and bill as one call.",
"family": "model-router",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/orcarouter/fusion-mini\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"orcarouter/fusion-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"orcarouter/fusion": {
"id": "orcarouter/fusion",
"name": "OrcaRouter Fusion",
"description": "Curated fan-out router that runs Claude Opus 4.8, GPT-5.5 and Gemini 3.1 Pro in parallel on hard requests, then has a Claude Opus 4.8 judge return the strongest single answer verbatim. Easy requests fall through to a cheaper default and bill as one call.",
"family": "model-router",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/orcarouter/fusion\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"orcarouter/fusion\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"orcarouter/fusion-flash": {
"id": "orcarouter/fusion-flash",
"name": "OrcaRouter Fusion Flash",
"description": "Budget Fusion panel that runs Gemini 3.5 Flash, MiniMax M2.7 and GLM 5.1 in parallel on hard requests, then has a Claude Opus 4.8 judge return the strongest single answer verbatim. Cost-sensitive fan-out over a 200K window.",
"family": "model-router",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/orcarouter/fusion-flash\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"orcarouter/fusion-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"orcarouter/auto": {
"id": "orcarouter/auto",
"name": "OrcaRouter Auto",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"family": "auto",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-01-01",
"last_updated": "2026-05-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/orcarouter/auto\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"orcarouter/auto\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7-highspeed": {
"id": "minimax/minimax-m2.7-highspeed",
"name": "MiniMax-M2.7-highspeed",
"description": "Low-latency M2.7 variant for interactive coding plans and agent loops",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/minimax/minimax-m2.7-highspeed\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7": {
"id": "minimax/minimax-m2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/minimax/minimax-m2.7\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5": {
"id": "minimax/minimax-m2.5",
"name": "MiniMax-M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/minimax/minimax-m2.5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m3": {
"id": "minimax/minimax-m3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 512000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/minimax/minimax-m3\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5-highspeed": {
"id": "minimax/minimax-m2.5-highspeed",
"name": "MiniMax-M2.5-highspeed",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-13",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.03,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/minimax/minimax-m2.5-highspeed\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.8": {
"id": "anthropic/claude-opus-4.8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/anthropic/claude-opus-4.8\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.7": {
"id": "anthropic/claude-opus-4.7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 30,
"output": 150,
"cache_read": 3,
"cache_write": 37.5
},
"provider": {
"body": {
"speed": "fast"
},
"headers": {
"anthropic-beta": "fast-mode-2026-02-01"
}
}
}
}
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/anthropic/claude-opus-4.7\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-5": {
"id": "anthropic/claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 10
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/anthropic/claude-opus-5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.6": {
"id": "anthropic/claude-sonnet-4.6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/anthropic/claude-sonnet-4.6\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-4.5": {
"id": "anthropic/claude-haiku-4.5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/anthropic/claude-haiku-4.5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.6": {
"id": "anthropic/claude-opus-4.6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 30,
"output": 150,
"cache_read": 3,
"cache_write": 37.5
},
"provider": {
"body": {
"speed": "fast"
},
"headers": {
"anthropic-beta": "fast-mode-2026-02-01"
}
}
}
}
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/anthropic/claude-opus-4.6\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5": {
"id": "anthropic/claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/anthropic/claude-fable-5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.5": {
"id": "anthropic/claude-sonnet-4.5",
"name": "Claude Sonnet 4.5 (latest)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/anthropic/claude-sonnet-4.5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.5": {
"id": "anthropic/claude-opus-4.5",
"name": "Claude Opus 4.5 (latest)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/anthropic/claude-opus-4.5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-5": {
"id": "anthropic/claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/anthropic/claude-sonnet-5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-26b-a4b-it": {
"id": "google/gemma-4-26b-a4b-it",
"name": "Gemma 4 26B A4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.06,
"output": 0.33,
"cache_read": 0.0075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-pro-preview-customtools": {
"id": "google/gemini-3.1-pro-preview-customtools",
"name": "Gemini 3.1 Pro Preview Custom Tools",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-3.1-pro-preview-customtools\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-pro-preview-customtools\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-pro-preview": {
"id": "google/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"input_audio": 2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash-lite": {
"id": "google/gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash-Lite",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01,
"input_audio": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.6-flash": {
"id": "google/gemini-3.6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 7.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-3.6-flash\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-flash-lite": {
"id": "google/gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"input_audio": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-flash": {
"id": "google/gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"cache_write": 0.08333,
"input_audio": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-3.5-flash\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-flash-lite-preview": {
"id": "google/gemini-3.1-flash-lite-preview",
"name": "Gemini 3.1 Flash Lite Preview",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"input_audio": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-3.1-flash-lite-preview\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-flash-lite-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-flash-lite": {
"id": "google/gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-robotics-er-1.6-preview": {
"id": "google/gemini-robotics-er-1.6-preview",
"name": "Gemini Robotics-ER 1.6 Preview",
"description": "Vision-language model for embodied reasoning: spatial understanding, task planning, and physical-world agentic robotics",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 0
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-14",
"last_updated": "2026-04-14",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 1,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-robotics-er-1.6-preview\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-robotics-er-1.6-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-flash-lite-latest": {
"id": "google/gemini-flash-lite-latest",
"name": "Gemini Flash-Lite Latest",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"input_audio": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-flash-lite-latest\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-flash-lite-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31b-it": {
"id": "google/gemma-4-31b-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.13,
"output": 0.38,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemma-4-31b-it\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3-flash-preview": {
"id": "google/gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"input_audio": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-3-flash-preview\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-pro": {
"id": "google/gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-2.5-pro\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-flash-latest": {
"id": "google/gemini-flash-latest",
"name": "Gemini Flash Latest",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-flash-latest\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-flash-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash": {
"id": "google/gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"input_audio": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/google/gemini-2.5-flash\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok/grok-4.3": {
"id": "grok/grok-4.3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/grok/grok-4.3\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"grok/grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok/grok-4.5": {
"id": "grok/grok-4.5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/grok/grok-4.5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"grok/grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok/grok-4.6": {
"id": "grok/grok-4.6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/grok/grok-4.6\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"grok/grok-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.2": {
"id": "meta/muse-spark-1.2",
"name": "Muse Spark 1.2",
"description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/meta/muse-spark-1.2\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.1": {
"id": "meta/muse-spark-1.1",
"name": "Muse Spark 1.1",
"description": "Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-08",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/meta/muse-spark-1.1\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-vision-exp": {
"id": "deepseek/deepseek-v4-flash-vision-exp",
"name": "DeepSeek V4 Flash Vision Exp",
"description": "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-21",
"last_updated": "2026-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.147,
"output": 0.295,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/deepseek/deepseek-v4-flash-vision-exp\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-vision-exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro-0813": {
"id": "deepseek/deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.442,
"output": 0.884,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/deepseek/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-0731": {
"id": "deepseek/deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.147,
"output": 0.295,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/deepseek/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-free": {
"id": "deepseek/deepseek-v4-flash-free",
"name": "DeepSeek V4 Flash (free)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/deepseek/deepseek-v4-flash-free\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash": {
"id": "deepseek/deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.147,
"output": 0.295,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/deepseek/deepseek-v4-flash\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-reasoner": {
"id": "deepseek/deepseek-reasoner",
"name": "DeepSeek Reasoner",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-12-01",
"last_updated": "2026-02-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.147,
"output": 0.295,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/deepseek/deepseek-reasoner\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-reasoner\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-chat": {
"id": "deepseek/deepseek-chat",
"name": "DeepSeek Chat",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-12-01",
"last_updated": "2026-02-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.147,
"output": 0.295,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/deepseek/deepseek-chat\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-chat\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro": {
"id": "deepseek/deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.442,
"output": 0.884,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/deepseek/deepseek-v4-pro\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-nano": {
"id": "openai/gpt-5-nano",
"name": "GPT-5 Nano",
"description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5-nano\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-nano": {
"id": "openai/gpt-4.1-nano",
"name": "GPT-4.1 nano",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-4.1-nano\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-2024-05-13": {
"id": "openai/gpt-4o-2024-05-13",
"name": "GPT-4o (2024-05-13)",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-05-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 5,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-4o-2024-05-13\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-2024-05-13\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-pro": {
"id": "openai/gpt-5-pro",
"name": "GPT-5 Pro",
"description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 272000
},
"cost": {
"input": 15,
"output": 120
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5-pro\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex-mini": {
"id": "openai/gpt-5.1-codex-mini",
"name": "GPT-5.1 Codex mini",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.1-codex-mini\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex": {
"id": "openai/gpt-5.1-codex",
"name": "GPT-5.1 Codex",
"description": "Codex GPT for repository edits, code review, and practical software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.1-codex\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-sol": {
"id": "openai/gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 4,
"output": 20,
"cache_read": 0.4,
"cache_write": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.6-sol\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-2024-08-06": {
"id": "openai/gpt-4o-2024-08-06",
"name": "GPT-4o (2024-08-06)",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-08-06",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-4o-2024-08-06\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-2024-08-06\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-codex": {
"id": "openai/gpt-5.2-codex",
"name": "GPT-5.2 Codex",
"description": "Code-specialist GPT for repository edits, reviews, and long-running software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.2-codex\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-pro": {
"id": "openai/gpt-5.2-pro",
"name": "GPT-5.2 Pro",
"description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 21,
"output": 168
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.2-pro\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-mini": {
"id": "openai/gpt-4.1-mini",
"name": "GPT-4.1 mini",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-4.1-mini\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-chat-latest": {
"id": "openai/gpt-5-chat-latest",
"name": "GPT-5 Chat (latest)",
"description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 100000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5-chat-latest\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-chat-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4": {
"id": "openai/gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"provider": {
"body": {
"service_tier": "priority"
}
}
}
}
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.4\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4-turbo": {
"id": "openai/gpt-4-turbo",
"name": "GPT-4 Turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 10,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-4-turbo\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1": {
"id": "openai/gpt-5.1",
"name": "GPT-5.1",
"description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.1\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-chat-latest": {
"id": "openai/gpt-5.1-chat-latest",
"name": "GPT-5.1 Chat",
"description": "Chat-tuned GPT-5.1 for polished assistants, writing, and product conversations",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.1-chat-latest\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-chat-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o": {
"id": "openai/gpt-4o",
"name": "GPT-4o",
"description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-4o\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-luna": {
"id": "openai/gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.6-luna\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-codex": {
"id": "openai/gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.3-codex\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini": {
"id": "openai/gpt-4o-mini",
"name": "GPT-4o mini",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-4o-mini\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1": {
"id": "openai/gpt-4.1",
"name": "GPT-4.1",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-4.1\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-nano": {
"id": "openai/gpt-5.4-nano",
"name": "GPT-5.4 nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.4-nano\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5-pro": {
"id": "openai/gpt-5.5-pro",
"name": "GPT-5.5 Pro",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 100000
},
"cost": {
"input": 30,
"output": 180,
"tiers": [
{
"input": 60,
"output": 270,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 60,
"output": 270
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.5-pro\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-mini": {
"id": "openai/gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15
},
"provider": {
"body": {
"service_tier": "priority"
}
}
}
}
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.4-mini\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-3.5-turbo": {
"id": "openai/gpt-3.5-turbo",
"name": "GPT-3.5-turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2021-09-01",
"release_date": "2023-03-01",
"last_updated": "2023-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16385,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 1.5,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-3.5-turbo\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-3.5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-mini": {
"id": "openai/gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5-mini\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.03,
"output": 0.17
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-oss-120b\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-pro": {
"id": "openai/gpt-5.4-pro",
"name": "GPT-5.4 Pro",
"description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180,
"tiers": [
{
"input": 60,
"output": 270,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 60,
"output": 270
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.4-pro\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-terra": {
"id": "openai/gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.6-terra\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4": {
"id": "openai/gpt-4",
"name": "GPT-4",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-11",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 30,
"output": 60
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-4\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2": {
"id": "openai/gpt-5.2",
"name": "GPT-5.2",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.2\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5": {
"id": "openai/gpt-5",
"name": "GPT-5",
"description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-chat-latest": {
"id": "openai/gpt-5.2-chat-latest",
"name": "GPT-5.2 Chat",
"description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.2-chat-latest\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-chat-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 12.5,
"output": 75,
"cache_read": 1.25
},
"provider": {
"body": {
"service_tier": "priority"
}
}
}
}
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-5.5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-2024-11-20": {
"id": "openai/gpt-4o-2024-11-20",
"name": "GPT-4o (2024-11-20)",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-11-20",
"last_updated": "2024-11-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/openai/gpt-4o-2024-11-20\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-2024-11-20\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi/kimi-k2.6": {
"id": "kimi/kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/kimi/kimi-k2.6\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"kimi/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi/kimi-k2.7-code": {
"id": "kimi/kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/kimi/kimi-k2.7-code\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"kimi/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi/kimi-k3": {
"id": "kimi/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3.3,
"output": 16.5,
"cache_read": 0.33
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/kimi/kimi-k3\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"kimi/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi/kimi-k2.5": {
"id": "kimi/kimi-k2.5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.1,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/kimi/kimi-k2.5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"kimi/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/hy3": {
"id": "tencent/hy3",
"name": "Hy3",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 192000,
"output": 128000
},
"cost": {
"input": 0.18,
"output": 0.59,
"cache_read": 0.059
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/tencent/hy3\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"tencent/hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/hy3-free": {
"id": "tencent/hy3-free",
"name": "Hy3 (free)",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 192000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/tencent/hy3-free\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"tencent/hy3-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7": {
"id": "z-ai/glm-4.7",
"name": "GLM-4.7",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/z-ai/glm-4.7\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.5-air": {
"id": "z-ai/glm-4.5-air",
"name": "GLM-4.5-Air",
"description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
"family": "glm-air",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0.2,
"output": 1.1,
"cache_read": 0.03,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/z-ai/glm-4.5-air\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.5-air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.6": {
"id": "z-ai/glm-4.6",
"name": "GLM-4.6",
"description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/z-ai/glm-4.6\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.2": {
"id": "z-ai/glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/z-ai/glm-5.2\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.3-flash": {
"id": "z-ai/glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.075,
"output": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/z-ai/glm-5.3-flash\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.5": {
"id": "z-ai/glm-4.5",
"name": "GLM-4.5",
"description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/z-ai/glm-4.5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5": {
"id": "z-ai/glm-5",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.26,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/z-ai/glm-5\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.1": {
"id": "z-ai/glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/z-ai/glm-5.1\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.3": {
"id": "z-ai/glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.26,
"output": 3.96,
"cache_read": 0.234
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/z-ai/glm-5.3\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.3-flash-free": {
"id": "z-ai/glm-5.3-flash-free",
"name": "GLM-5.3-Flash (free)",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"orcarouter/z-ai/glm-5.3-flash-free\", apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.orcarouter.ai/v1\")!,\n apiKey: processEnvironment[\"ORCAROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3-flash-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"routing-run": {
"id": "routing-run",
"name": "routing.run",
"baseURL": "https://api.routing.run/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ROUTING_RUN_API_KEY"
],
"doc": "https://docs.routing.run/api-reference/models",
"modelCount": 15,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/claude-sonnet-4-6\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-9b": {
"id": "qwen3.5-9b",
"name": "Qwen3.5 9B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32000
},
"cost": {
"input": 0.16,
"output": 0.48
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/qwen3.5-9b\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-sol": {
"id": "gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/gpt-5.6-sol\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 0.275,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/kimi-k2.6\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 0.8,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/glm-5.2\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.112,
"output": 0.224
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/deepseek-v4-flash\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6-nitro": {
"id": "kimi-k2.6-nitro",
"name": "Kimi K2.6 Nitro",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 0.275,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/kimi-k2.6-nitro\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6-nitro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code": {
"id": "kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 0.275,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/kimi-k2.7-code\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-ultra": {
"id": "nemotron-3-ultra",
"name": "Nemotron 3 Ultra 550B A55B",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32000
},
"cost": {
"input": 0.1,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/nemotron-3-ultra\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-nitro": {
"id": "glm-5.2-nitro",
"name": "GLM 5.2 Nitro",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 0.8,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/glm-5.2-nitro\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-nitro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-luna": {
"id": "gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.7,
"output": 4.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/gpt-5.6-luna\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32000
},
"cost": {
"input": 5,
"output": 25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/claude-opus-4-8\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.348,
"output": 0.696
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/deepseek-v4-pro\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code-nitro": {
"id": "kimi-k2.7-code-nitro",
"name": "Kimi K2.7 Code Nitro",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 0.275,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/kimi-k2.7-code-nitro\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code-nitro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-terra": {
"id": "gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 1.5,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"routing-run/gpt-5.6-terra\", apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.routing.run/v1\")!,\n apiKey: processEnvironment[\"ROUTING_RUN_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"llmtech": {
"id": "llmtech",
"name": "LLM Tech",
"baseURL": "https://api.llmtech.eu/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"LLMTECH_API_KEY"
],
"doc": "https://llmtech.eu/models/qwen3.8-27b",
"modelCount": 1,
"models": {
"unsloth/Qwen3.8-27B-NVFP4": {
"id": "unsloth/Qwen3.8-27B-NVFP4",
"name": "Qwen3.8 27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.25,
"output": 2.09,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtech/unsloth/Qwen3.8-27B-NVFP4\", apiKey: processEnvironment[\"LLMTECH_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmtech.eu/v1\")!,\n apiKey: processEnvironment[\"LLMTECH_API_KEY\"]\n)\nlet session = provider.model(\"unsloth/Qwen3.8-27B-NVFP4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"sap-ai-core": {
"id": "sap-ai-core",
"name": "SAP AI Core",
"baseURL": "",
"npm": "@jerome-benoit/sap-ai-provider-v2",
"swiftDriver": "openaiChat",
"env": [
"AICORE_SERVICE_KEY"
],
"doc": "https://help.sap.com/docs/sap-ai-core",
"modelCount": 49,
"models": {
"gpt-5-nano": {
"id": "gpt-5-nano",
"name": "gpt-5-nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-5-nano\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-nano": {
"id": "gpt-4.1-nano",
"name": "gpt-4.1-nano",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.08,
"output": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-4.1-nano\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-4.5-sonnet": {
"id": "anthropic--claude-4.5-sonnet",
"name": "anthropic--claude-4.5-sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-4.5-sonnet\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-4.5-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai--mistral-medium": {
"id": "mistralai--mistral-medium",
"name": "Mistral Medium 3.5",
"description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
"family": "mistral-medium",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-29",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/mistralai--mistral-medium\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"mistralai--mistral-medium\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-sol": {
"id": "gpt-5.6-sol",
"name": "gpt-5.6-sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-5.6-sol\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-4.5-haiku": {
"id": "anthropic--claude-4.5-haiku",
"name": "anthropic--claude-4.5-haiku",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-4.5-haiku\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-4.5-haiku\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite": {
"id": "gemini-2.5-flash-lite",
"name": "gemini-2.5-flash-lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 512,
"max": 24576
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01,
"input_audio": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cohere--command-a-reasoning": {
"id": "cohere--command-a-reasoning",
"name": "cohere--command-a-reasoning",
"description": "Cohere reasoning model for multilingual enterprise agents, tools, and complex workflows",
"family": "command-a",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-06-01",
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0.63,
"output": 5.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/cohere--command-a-reasoning\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"cohere--command-a-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-mini": {
"id": "gpt-4.1-mini",
"name": "gpt-4.1-mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-4.1-mini\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "gpt-5.4",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-5.4\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-4.8-opus": {
"id": "anthropic--claude-4.8-opus",
"name": "anthropic--claude-4.8-opus",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-4.8-opus\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-4.8-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite": {
"id": "gemini-3.1-flash-lite",
"name": "gemini-3.1-flash-lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"input_audio": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash": {
"id": "gemini-3.5-flash",
"name": "gemini-3.5-flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"input_audio": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gemini-3.5-flash\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-luna": {
"id": "gpt-5.6-luna",
"name": "gpt-5.6-luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 1,
"output": 6,
"cache_read": 0.1,
"tiers": [
{
"input": 2,
"output": 9,
"cache_read": 0.2,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 2,
"output": 9,
"cache_read": 0.2
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-5.6-luna\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon--titan-embed-text": {
"id": "amazon--titan-embed-text",
"name": "amazon--titan-embed-text",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-04-30",
"last_updated": "2024-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"cost": {
"input": 0.14,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/amazon--titan-embed-text\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"amazon--titan-embed-text\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-4.5-opus": {
"id": "anthropic--claude-4.5-opus",
"name": "anthropic--claude-4.5-opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-4.5-opus\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-4.5-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-3.5-sonnet": {
"id": "anthropic--claude-3.5-sonnet",
"name": "anthropic--claude-3.5-sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04-30",
"release_date": "2024-10-22",
"last_updated": "2024-10-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"status": "deprecated",
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-3.5-sonnet\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-3.5-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai--mistral-medium-instruct": {
"id": "mistralai--mistral-medium-instruct",
"name": "mistralai--mistral-medium-instruct",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-medium",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-05-07",
"last_updated": "2025-05-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.36,
"output": 1.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/mistralai--mistral-medium-instruct\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"mistralai--mistral-medium-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash-lite": {
"id": "gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1": {
"id": "gpt-4.1",
"name": "gpt-4.1",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.32
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-4.1\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-4.6-opus": {
"id": "anthropic--claude-4.6-opus",
"name": "anthropic--claude-4.6-opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-4.6-opus\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-4.6-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar": {
"id": "sonar",
"name": "sonar",
"description": "Sonar search model for current answers, retrieval, and citation-backed chat",
"family": "sonar",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 1,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/sonar\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"sonar\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-4-sonnet": {
"id": "anthropic--claude-4-sonnet",
"name": "anthropic--claude-4-sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-4-sonnet\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-4-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai--mistral-small": {
"id": "mistralai--mistral-small",
"name": "mistralai--mistral-small",
"description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents",
"family": "mistral-small",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.07,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/mistralai--mistral-small\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"mistralai--mistral-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon--nova-pro": {
"id": "amazon--nova-pro",
"name": "amazon--nova-pro",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "nova-pro",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-12-03",
"last_updated": "2024-12-03",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 300000,
"output": 8192
},
"cost": {
"input": 0.56,
"output": 2.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/amazon--nova-pro\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"amazon--nova-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-3-opus": {
"id": "anthropic--claude-3-opus",
"name": "anthropic--claude-3-opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-08-31",
"release_date": "2024-02-29",
"last_updated": "2024-02-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 4096
},
"status": "deprecated",
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-3-opus\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-3-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia--llama-3.2-nv-embedqa-1b": {
"id": "nvidia--llama-3.2-nv-embedqa-1b",
"name": "nvidia--llama-3.2-nv-embedqa-1b",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-09-25",
"last_updated": "2024-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 4096
},
"cost": {
"input": 0.07,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/nvidia--llama-3.2-nv-embedqa-1b\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"nvidia--llama-3.2-nv-embedqa-1b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-4.7-opus": {
"id": "anthropic--claude-4.7-opus",
"name": "anthropic--claude-4.7-opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-4.7-opus\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-4.7-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-embedding-2": {
"id": "gemini-embedding-2",
"name": "Gemini Embedding 2",
"description": "Multimodal embedding model mapping text, images, video, audio, and PDFs into a unified embedding space",
"family": "gemini",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2025-11",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 3072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gemini-embedding-2\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gemini-embedding-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sap-abap-1": {
"id": "sap-abap-1",
"name": "sap-abap-1",
"description": "SAP-hosted model for ABAP code generation and enterprise development tasks",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-11-26",
"last_updated": "2025-11-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 4096
},
"cost": {
"input": 0.48,
"output": 1.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/sap-abap-1\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"sap-abap-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon--nova-lite": {
"id": "amazon--nova-lite",
"name": "amazon--nova-lite",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "nova-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.3,
"output": 2.37
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/amazon--nova-lite\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"amazon--nova-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-3-haiku": {
"id": "anthropic--claude-3-haiku",
"name": "anthropic--claude-3-haiku",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-08-31",
"release_date": "2024-03-13",
"last_updated": "2024-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 4096
},
"cost": {
"input": 0.25,
"output": 1.25,
"cache_read": 0.03,
"cache_write": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-3-haiku\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-3-haiku\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar-pro": {
"id": "sonar-pro",
"name": "sonar-pro",
"description": "Advanced Sonar search model for deeper research and cited synthesis",
"family": "sonar-pro",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 3,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/sonar-pro\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"sonar-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"text-embedding-3-small": {
"id": "text-embedding-3-small",
"name": "text-embedding-3-small",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-01-25",
"last_updated": "2024-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8191,
"output": 1536
},
"cost": {
"input": 0.02,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/text-embedding-3-small\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"text-embedding-3-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon--nova-micro": {
"id": "amazon--nova-micro",
"name": "amazon--nova-micro",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "nova-micro",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-12-03",
"last_updated": "2024-12-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.03,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/amazon--nova-micro\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"amazon--nova-micro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-mini": {
"id": "gpt-5-mini",
"name": "gpt-5-mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-5-mini\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar-deep-research": {
"id": "sonar-deep-research",
"name": "sonar-deep-research",
"description": "Sonar search model for current answers, retrieval, and citation-backed chat",
"family": "sonar-deep-research",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-02-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"reasoning": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/sonar-deep-research\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"sonar-deep-research\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"text-embedding-3-large": {
"id": "text-embedding-3-large",
"name": "text-embedding-3-large",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-01-25",
"last_updated": "2024-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8191,
"output": 3072
},
"cost": {
"input": 0.09,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/text-embedding-3-large\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"text-embedding-3-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "gemini-2.5-pro",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-03-25",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gemini-2.5-pro\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-terra": {
"id": "gpt-5.6-terra",
"name": "gpt-5.6-terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-5.6-terra\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2": {
"id": "gpt-5.2",
"name": "gpt-5.2",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 9.44,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-5.2\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-4.6-sonnet": {
"id": "anthropic--claude-4.6-sonnet",
"name": "anthropic--claude-4.6-sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-4.6-sonnet\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-4.6-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5": {
"id": "gpt-5",
"name": "gpt-5",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-5\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "gemini-2.5-flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-04-17",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"input_audio": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gemini-2.5-flash\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-4-opus": {
"id": "anthropic--claude-4-opus",
"name": "anthropic--claude-4-opus",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"status": "deprecated",
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-4-opus\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-4-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "gpt-5.5",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gpt-5.5\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-3-sonnet": {
"id": "anthropic--claude-3-sonnet",
"name": "anthropic--claude-3-sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-08-31",
"release_date": "2024-03-04",
"last_updated": "2024-03-04",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 4096
},
"status": "deprecated",
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-3-sonnet\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-3-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-embedding": {
"id": "gemini-embedding",
"name": "Gemini Embedding 001",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "gemini",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2025-05",
"release_date": "2025-05-20",
"last_updated": "2025-05-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2048,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/gemini-embedding\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"gemini-embedding\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic--claude-3.7-sonnet": {
"id": "anthropic--claude-3.7-sonnet",
"name": "anthropic--claude-3.7-sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10-31",
"release_date": "2025-02-24",
"last_updated": "2025-02-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"status": "deprecated",
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sap-ai-core/anthropic--claude-3.7-sonnet\", apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AICORE_SERVICE_KEY\"]\n)\nlet session = provider.model(\"anthropic--claude-3.7-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"alibaba-coding-plan-cn": {
"id": "alibaba-coding-plan-cn",
"name": "Alibaba Coding Plan (China)",
"baseURL": "https://coding.dashscope.aliyuncs.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ALIBABA_CODING_PLAN_API_KEY"
],
"doc": "https://help.aliyun.com/zh/model-studio/coding-plan",
"modelCount": 12,
"models": {
"glm-4.7": {
"id": "glm-4.7",
"name": "GLM-4.7",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 16384
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/glm-4.7\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-max": {
"id": "qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.5,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/qwen3.7-max\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-plus": {
"id": "qwen3-coder-plus",
"name": "Qwen3 Coder Plus",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/qwen3-coder-plus\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-plus": {
"id": "qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/qwen3.6-plus\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-next": {
"id": "qwen3-coder-next",
"name": "Qwen3 Coder Next",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-03",
"last_updated": "2026-02-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/qwen3-coder-next\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.5": {
"id": "MiniMax-M2.5",
"name": "MiniMax-M2.5",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 24576
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/MiniMax-M2.5\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-flash": {
"id": "qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.1875,
"output": 1.125,
"cache_write": 0.234375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/qwen3.6-flash\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5": {
"id": "glm-5",
"name": "GLM-5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-11",
"last_updated": "2026-02-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 202752,
"output": 16384
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/glm-5\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "Kimi K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/kimi-k2.5\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-plus": {
"id": "qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/qwen3.7-plus\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-max-2026-01-23": {
"id": "qwen3-max-2026-01-23",
"name": "Qwen3 Max",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-23",
"last_updated": "2026-01-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/qwen3-max-2026-01-23\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-max-2026-01-23\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-plus": {
"id": "qwen3.5-plus",
"name": "Qwen3.5 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-coding-plan-cn/qwen3.5-plus\", apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://coding.dashscope.aliyuncs.com/v1\")!,\n apiKey: processEnvironment[\"ALIBABA_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"azure-cognitive-services": {
"id": "azure-cognitive-services",
"name": "Azure Cognitive Services",
"baseURL": "",
"npm": "@ai-sdk/azure",
"swiftDriver": "openaiChat",
"env": [
"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME",
"AZURE_COGNITIVE_SERVICES_API_KEY"
],
"doc": "https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models",
"modelCount": 74,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-sonnet-4-6\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-sol": {
"id": "gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.6-sol\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5": {
"id": "claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-opus-5\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/models",
"shape": "completions"
},
"cost": {
"input": 0.95,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/kimi-k2.6\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5": {
"id": "claude-opus-4-5",
"name": "Claude Opus 4.5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-11-24",
"last_updated": "2025-08-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-opus-4-5\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-opus-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.4\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5-1": {
"id": "claude-fable-5-1",
"name": "Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-fable-5-1\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-fable-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-07-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-opus-4-6\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-luna": {
"id": "gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 1,
"output": 6,
"cache_read": 0.1,
"cache_write": 1.25,
"tiers": [
{
"input": 2,
"output": 9,
"cache_read": 0.2,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 2,
"output": 9,
"cache_read": 0.2,
"cache_write": 2.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.6-luna\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7": {
"id": "claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-opus-4-7\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"status": "beta",
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-fable-5\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-nano": {
"id": "gpt-5.4-nano",
"name": "GPT-5.4 Nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.4-nano\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-chat-latest": {
"id": "gpt-chat-latest",
"name": "GPT Chat Latest",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-05-05",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 111616,
"output": 16384
},
"status": "beta",
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-chat-latest\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-chat-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-mini": {
"id": "gpt-5.4-mini",
"name": "GPT-5.4 Mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.4-mini\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5": {
"id": "claude-haiku-4-5",
"name": "Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-11-18",
"last_updated": "2025-11-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-haiku-4-5\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5": {
"id": "claude-sonnet-4-5",
"name": "Claude Sonnet 4.5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-11-18",
"last_updated": "2025-11-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-sonnet-4-5\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1": {
"id": "claude-opus-4-1",
"name": "Claude Opus 4.1",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-11-18",
"last_updated": "2025-11-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-opus-4-1\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-opus-4-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/models",
"shape": "completions"
},
"cost": {
"input": 0.6,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/kimi-k2.5\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2025-12-31",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-opus-4-8\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-pro": {
"id": "gpt-5.4-pro",
"name": "GPT-5.4 Pro",
"description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180,
"tiers": [
{
"input": 60,
"output": 270,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 60,
"output": 270
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.4-pro\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-terra": {
"id": "gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"cache_write": 3.125,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"cache_write": 6.25,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"cache_write": 6.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.6-terra\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5": {
"id": "claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-sonnet-5\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-mythos-5": {
"id": "claude-mythos-5",
"name": "Claude Mythos 5",
"description": "Restricted Claude model for advanced cybersecurity and biology research workflows",
"family": "claude-mythos",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"status": "beta",
"provider": {
"npm": "@ai-sdk/anthropic",
"api": "https://${AZURE_COGNITIVE_SERVICES_RESOURCE_NAME}.services.ai.azure.com/anthropic/v1"
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/claude-mythos-5\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"claude-mythos-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.5\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o3": {
"id": "o3",
"name": "o3",
"description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/o3\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o3-mini": {
"id": "o3-mini",
"name": "o3-mini",
"description": "Smaller o-series reasoner for economical coding, math, and planning tasks",
"family": "o-mini",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2024-12-20",
"last_updated": "2025-01-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"status": "deprecated",
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/o3-mini\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o4-mini": {
"id": "o4-mini",
"name": "o4-mini",
"description": "Fast o-series model for compact reasoning, coding, and tool use",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"status": "deprecated",
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/o4-mini\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"o4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-70b-instruct": {
"id": "llama-3.3-70b-instruct",
"name": "Llama-3.3-70B-Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.71,
"output": 0.71
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5": {
"id": "gpt-5",
"name": "GPT-5",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2": {
"id": "gpt-5.2",
"name": "GPT-5.2",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.2\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-3.5-turbo-instruct": {
"id": "gpt-3.5-turbo-instruct",
"name": "GPT-3.5 Turbo Instruct",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2021-08",
"release_date": "2023-09-21",
"last_updated": "2023-09-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 4096,
"output": 4096
},
"status": "deprecated",
"cost": {
"input": 1.5,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-3.5-turbo-instruct\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-3.5-turbo-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"codestral-2501": {
"id": "codestral-2501",
"name": "Codestral 25.01",
"description": "Mistral coding model for code completion, generation, and developer workflows",
"family": "codestral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-03",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.3,
"output": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/codestral-2501\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"codestral-2501\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"text-embedding-3-large": {
"id": "text-embedding-3-large",
"name": "text-embedding-3-large",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2024-01-25",
"last_updated": "2024-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8191,
"output": 3072
},
"cost": {
"input": 0.13,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/text-embedding-3-large\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"text-embedding-3-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-mini": {
"id": "gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5-mini\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"text-embedding-3-small": {
"id": "text-embedding-3-small",
"name": "text-embedding-3-small",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2024-01-25",
"last_updated": "2024-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8191,
"output": 1536
},
"cost": {
"input": 0.02,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/text-embedding-3-small\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"text-embedding-3-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-medium-2505": {
"id": "mistral-medium-2505",
"name": "Mistral Medium 3",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-medium",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-05-07",
"last_updated": "2025-05-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.4,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/mistral-medium-2505\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"mistral-medium-2505\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"phi-4-reasoning": {
"id": "phi-4-reasoning",
"name": "Phi-4-reasoning",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"family": "phi",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2023-10",
"release_date": "2024-12-11",
"last_updated": "2024-12-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 4096
},
"cost": {
"input": 0.125,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/phi-4-reasoning\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"phi-4-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3.2-speciale": {
"id": "deepseek-v3.2-speciale",
"name": "DeepSeek-V3.2-Speciale",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.58,
"output": 1.68
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/deepseek-v3.2-speciale\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"deepseek-v3.2-speciale\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cohere-embed-v3-multilingual": {
"id": "cohere-embed-v3-multilingual",
"name": "Embed v3 Multilingual",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "cohere-embed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2023-11-07",
"last_updated": "2023-11-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512,
"output": 1024
},
"cost": {
"input": 0.1,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/cohere-embed-v3-multilingual\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"cohere-embed-v3-multilingual\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"model-router": {
"id": "model-router",
"name": "Model Router",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"family": "model-router",
"attachment": true,
"reasoning": false,
"tool_call": true,
"release_date": "2025-05-19",
"last_updated": "2025-11-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 16384
},
"cost": {
"input": 0.14,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/model-router\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"model-router\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"text-embedding-ada-002": {
"id": "text-embedding-ada-002",
"name": "text-embedding-ada-002",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2022-12-15",
"last_updated": "2022-12-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"cost": {
"input": 0.1,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/text-embedding-ada-002\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"text-embedding-ada-002\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1": {
"id": "gpt-4.1",
"name": "GPT-4.1",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"status": "deprecated",
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-4.1\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o-mini": {
"id": "gpt-4o-mini",
"name": "GPT-4o mini",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"status": "deprecated",
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-4o-mini\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-4-scout-17b-16e-instruct": {
"id": "llama-4-scout-17b-16e-instruct",
"name": "Llama 4 Scout 17B 16E Instruct",
"description": "Open multimodal Llama model for long-context analysis and efficient agents",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.2,
"output": 0.78
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/llama-4-scout-17b-16e-instruct\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"llama-4-scout-17b-16e-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.3-codex": {
"id": "gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-02-24",
"last_updated": "2026-02-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.3-codex\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4-turbo-vision": {
"id": "gpt-4-turbo-vision",
"name": "GPT-4 Turbo Vision",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-11",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"status": "deprecated",
"cost": {
"input": 10,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-4-turbo-vision\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-4-turbo-vision\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3.2": {
"id": "deepseek-v3.2",
"name": "DeepSeek-V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.58,
"output": 1.68
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/deepseek-v3.2\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o": {
"id": "gpt-4o",
"name": "GPT-4o",
"description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"status": "deprecated",
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-4o\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o1": {
"id": "o1",
"name": "o1",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2023-09",
"release_date": "2024-12-05",
"last_updated": "2024-12-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"status": "deprecated",
"cost": {
"input": 15,
"output": 60,
"cache_read": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/o1\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"o1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1": {
"id": "deepseek-r1",
"name": "DeepSeek-R1",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-01-20",
"last_updated": "2025-01-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"status": "deprecated",
"cost": {
"input": 1.35,
"output": 5.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/deepseek-r1\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"deepseek-r1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1": {
"id": "gpt-5.1",
"name": "GPT-5.1",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-14",
"last_updated": "2025-11-14",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text",
"image",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.1\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4-turbo": {
"id": "gpt-4-turbo",
"name": "GPT-4 Turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"status": "deprecated",
"cost": {
"input": 10,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-4-turbo\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-4-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"phi-4-mini": {
"id": "phi-4-mini",
"name": "Phi-4-mini",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "phi",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-10",
"release_date": "2024-12-11",
"last_updated": "2024-12-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.075,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/phi-4-mini\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"phi-4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"phi-4-multimodal": {
"id": "phi-4-multimodal",
"name": "Phi-4-multimodal",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"family": "phi",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2023-10",
"release_date": "2024-12-11",
"last_updated": "2024-12-11",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.08,
"output": 0.32,
"input_audio": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/phi-4-multimodal\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"phi-4-multimodal\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-3.5-turbo-0125": {
"id": "gpt-3.5-turbo-0125",
"name": "GPT-3.5 Turbo 0125",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2021-08",
"release_date": "2024-01-25",
"last_updated": "2024-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"output": 16384
},
"status": "deprecated",
"cost": {
"input": 0.5,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-3.5-turbo-0125\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-3.5-turbo-0125\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-mini": {
"id": "gpt-4.1-mini",
"name": "GPT-4.1 mini",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"status": "deprecated",
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-4.1-mini\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"phi-4": {
"id": "phi-4",
"name": "Phi-4",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"family": "phi",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2023-10",
"release_date": "2024-12-11",
"last_updated": "2024-12-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.125,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/phi-4\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"phi-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cohere-embed-v-4-0": {
"id": "cohere-embed-v-4-0",
"name": "Embed v4",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "cohere-embed",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-04-15",
"last_updated": "2025-04-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 1536
},
"cost": {
"input": 0.12,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/cohere-embed-v-4-0\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"cohere-embed-v-4-0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cohere-embed-v3-english": {
"id": "cohere-embed-v3-english",
"name": "Embed v3 English",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "cohere-embed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2023-11-07",
"last_updated": "2023-11-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512,
"output": 1024
},
"cost": {
"input": 0.1,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/cohere-embed-v3-english\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"cohere-embed-v3-english\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-4-maverick-17b-128e-instruct-fp8": {
"id": "llama-4-maverick-17b-128e-instruct-fp8",
"name": "Llama 4 Maverick 17B 128E Instruct FP8",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 16384
},
"cost": {
"input": 0.25,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/llama-4-maverick-17b-128e-instruct-fp8\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"llama-4-maverick-17b-128e-instruct-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2-codex": {
"id": "gpt-5.2-codex",
"name": "GPT-5.2 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-01-14",
"last_updated": "2026-01-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.2-codex\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.2-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cohere-command-a": {
"id": "cohere-command-a",
"name": "Command A",
"description": "Cohere command model for multilingual enterprise agents, tools, and chat",
"family": "command-a",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-06-01",
"release_date": "2025-03-13",
"last_updated": "2025-03-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 2.5,
"output": 10
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/cohere-command-a\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"cohere-command-a\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"phi-4-reasoning-plus": {
"id": "phi-4-reasoning-plus",
"name": "Phi-4-reasoning-plus",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"family": "phi",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2023-10",
"release_date": "2024-12-11",
"last_updated": "2024-12-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 4096
},
"cost": {
"input": 0.125,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/phi-4-reasoning-plus\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"phi-4-reasoning-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex": {
"id": "gpt-5.1-codex",
"name": "GPT-5.1 Codex",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "gpt-codex",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-14",
"last_updated": "2025-11-14",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text",
"image",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.1-codex\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex-mini": {
"id": "gpt-5.1-codex-mini",
"name": "GPT-5.1 Codex Mini",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-14",
"last_updated": "2025-11-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5.1-codex-mini\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5.1-codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"codex-mini": {
"id": "codex-mini",
"name": "Codex Mini",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2024-04",
"release_date": "2025-05-16",
"last_updated": "2025-05-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"status": "deprecated",
"cost": {
"input": 1.5,
"output": 6,
"cache_read": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/codex-mini\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"phi-4-mini-reasoning": {
"id": "phi-4-mini-reasoning",
"name": "Phi-4-mini-reasoning",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "phi",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2023-10",
"release_date": "2024-12-11",
"last_updated": "2024-12-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.075,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/phi-4-mini-reasoning\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"phi-4-mini-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-pro": {
"id": "gpt-5-pro",
"name": "GPT-5 Pro",
"description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 272000
},
"cost": {
"input": 15,
"output": 120
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5-pro\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-codex": {
"id": "gpt-5-codex",
"name": "GPT-5-Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-09-15",
"last_updated": "2025-09-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5-codex\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-3.5-turbo-1106": {
"id": "gpt-3.5-turbo-1106",
"name": "GPT-3.5 Turbo 1106",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2021-08",
"release_date": "2023-11-06",
"last_updated": "2023-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"output": 16384
},
"status": "deprecated",
"cost": {
"input": 1,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-3.5-turbo-1106\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-3.5-turbo-1106\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-nano": {
"id": "gpt-4.1-nano",
"name": "GPT-4.1 nano",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"status": "deprecated",
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-4.1-nano\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ministral-3b": {
"id": "ministral-3b",
"name": "Ministral 3B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-03",
"release_date": "2024-10-22",
"last_updated": "2024-10-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.04,
"output": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/ministral-3b\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"ministral-3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-2503": {
"id": "mistral-small-2503",
"name": "Mistral Small 3.1",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2025-03-01",
"last_updated": "2025-03-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/mistral-small-2503\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"mistral-small-2503\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-nano": {
"id": "gpt-5-nano",
"name": "GPT-5 Nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"azure-cognitive-services/gpt-5-nano\", apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AZURE_COGNITIVE_SERVICES_RESOURCE_NAME\"]\n)\nlet session = provider.model(\"gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"regolo-ai": {
"id": "regolo-ai",
"name": "Regolo AI",
"baseURL": "https://api.regolo.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"REGOLO_API_KEY"
],
"doc": "https://docs.regolo.ai/",
"modelCount": 18,
"models": {
"qwen3.5-9b": {
"id": "qwen3.5-9b",
"name": "Qwen3.5-9B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-01",
"last_updated": "2026-02-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 8192
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/qwen3.5-9b\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-27b": {
"id": "qwen3.8-27b",
"name": "Qwen3.8 27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 120000,
"output": 120000
},
"cost": {
"input": 0.58,
"output": 2.42
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/qwen3.8-27b\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"faster-whisper-large-v3": {
"id": "faster-whisper-large-v3",
"name": "Faster Whisper Large v3",
"description": "Open Whisper checkpoint for robust multilingual transcription and captioning",
"family": "whisper",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2024-10-01",
"last_updated": "2024-10-01",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 448,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/faster-whisper-large-v3\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"faster-whisper-large-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma4-31b": {
"id": "gemma4-31b",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 100000,
"output": 100000
},
"cost": {
"input": 0.46,
"output": 2.42
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/gemma4-31b\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"gemma4-31b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"brick-complexity-pro": {
"id": "brick-complexity-pro",
"name": "Brick Complexity Pro",
"description": "Complexity classifier that powers the Brick semantic router by extracting query difficulty",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2026-02-06",
"last_updated": "2026-02-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 100000,
"output": 15000
},
"cost": {
"input": 0.12,
"output": 0.46
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/brick-complexity-pro\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"brick-complexity-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"apertus-70b": {
"id": "apertus-70b",
"name": "Apertus 70B",
"description": "Fully open 70B multilingual LLM supporting 1800+ languages with 65K context. Trained on 15T tokens of compliant open data. Apache 2.0, EU AI Act compliant.",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-02",
"last_updated": "2025-09-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 30000,
"output": 30000
},
"cost": {
"input": 0.46,
"output": 2.42
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/apertus-70b\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"apertus-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-20b": {
"id": "gpt-oss-20b",
"name": "GPT-OSS-20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-01",
"last_updated": "2026-03-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.4,
"output": 1.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/gpt-oss-20b\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-next": {
"id": "qwen3-coder-next",
"name": "Qwen3-Coder-Next",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-01",
"last_updated": "2026-03-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/qwen3-coder-next\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-embedding-8b": {
"id": "qwen3-embedding-8b",
"name": "Qwen3-Embedding-8B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-02-01",
"last_updated": "2026-02-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0.1,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/qwen3-embedding-8b\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-embedding-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-reranker-4b": {
"id": "qwen3-reranker-4b",
"name": "Qwen3-Reranker-4B",
"description": "Reranking model for improving retrieval quality in search and recommendation systems",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-02-01",
"last_updated": "2026-02-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0.12,
"output": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/qwen3-reranker-4b\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-reranker-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm5.2": {
"id": "glm5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 96000,
"output": 96000
},
"cost": {
"input": 2.31,
"output": 6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/glm5.2\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"glm5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"brick-v1-beta": {
"id": "brick-v1-beta",
"name": "Brick v1 Beta",
"description": "Semantic router by Regolo.ai that directs each request to the most suitable model, optimizing costs and performance",
"family": "model-router",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2026-02-06",
"last_updated": "2026-02-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 100000,
"output": 15000
},
"status": "beta",
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/brick-v1-beta\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"brick-v1-beta\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-image": {
"id": "qwen-image",
"name": "Qwen-Image",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-03-01",
"last_updated": "2026-03-01",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/qwen-image\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"qwen-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT-OSS-120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 1,
"output": 4.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/gpt-oss-120b\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ocr-2": {
"id": "deepseek-ocr-2",
"name": "DeepSeek OCR 2",
"description": "High-accuracy OCR model for extracting text from documents, screenshots, receipts, and natural scenes",
"attachment": true,
"reasoning": false,
"tool_call": false,
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 4000,
"output": 4000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/deepseek-ocr-2\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ocr-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-4-119b": {
"id": "mistral-small-4-119b",
"name": "Mistral Small 4 119B",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-15",
"last_updated": "2026-03-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0.75,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/mistral-small-4-119b\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small-4-119b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-70b-instruct": {
"id": "llama-3.3-70b-instruct",
"name": "Llama 3.3 70B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-04-28",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.6,
"output": 2.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-122b": {
"id": "qwen3.5-122b",
"name": "Qwen3.5-122B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-01",
"last_updated": "2026-02-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.9,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"regolo-ai/qwen3.5-122b\", apiKey: processEnvironment[\"REGOLO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.regolo.ai/v1\")!,\n apiKey: processEnvironment[\"REGOLO_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-122b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"kenari": {
"id": "kenari",
"name": "Kenari",
"baseURL": "https://kenari.id/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"KENARI_API_KEY"
],
"doc": "https://kenari.id/docs",
"modelCount": 59,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/claude-sonnet-4-6\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-6-terra": {
"id": "gpt-5-6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gpt-5-6-terra\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5-1": {
"id": "glm-5-1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/glm-5-1\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"glm-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-7-flash": {
"id": "gemini-3-7-flash",
"name": "Gemini 3.7 Flash",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gemini-3-7-flash\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2-5-flash": {
"id": "gemini-2-5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gemini-2-5-flash\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2-5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2-7-highspeed": {
"id": "minimax-m2-7-highspeed",
"name": "MiniMax-M2.7-highspeed",
"description": "Low-latency M2.7 variant for interactive coding plans and agent loops",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/minimax-m2-7-highspeed\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2-7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-7-code:free": {
"id": "kimi-k2-7-code:free",
"name": "Kimi K2.7 Code (Free)",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/kimi-k2-7-code:free\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-7-code:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-7-plus": {
"id": "qwen3-7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/qwen3-7-plus\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-5-flash": {
"id": "gemini-3-5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gemini-3-5-flash\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5": {
"id": "claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/claude-opus-5\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash:free": {
"id": "deepseek-v4-flash:free",
"name": "DeepSeek V4 Flash (Free)",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/deepseek-v4-flash:free\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-6-sol": {
"id": "gpt-5-6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gpt-5-6-sol\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-6-luna": {
"id": "gpt-5-6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gpt-5-6-luna\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m3": {
"id": "minimax-m3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 512000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/minimax-m3\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/deepseek-v4-flash\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-medium-3-5:free": {
"id": "mistral-medium-3-5:free",
"name": "Mistral Medium 3.5 (Free)",
"description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
"family": "mistral-medium",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-29",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/mistral-medium-3-5:free\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"mistral-medium-3-5:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-6-flash": {
"id": "gemini-3-6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gemini-3-6-flash\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5-3": {
"id": "glm-5-3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/glm-5-3\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"glm-5-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3-7-flash:free": {
"id": "step-3-7-flash:free",
"name": "Step 3.7 Flash (Free)",
"description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/step-3-7-flash:free\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"step-3-7-flash:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-7-code": {
"id": "kimi-k2-7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/kimi-k2-7-code\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-20b": {
"id": "gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open-weight GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gpt-oss-20b\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-1-pro": {
"id": "gemini-3-1-pro",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gemini-3-1-pro\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-1-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"hy3": {
"id": "hy3",
"name": "Hy3",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 192000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/hy3\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-imagine-image-2-0": {
"id": "grok-imagine-image-2-0",
"name": "Grok Imagine Image 2.0",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-08-07",
"last_updated": "2026-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 8000,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/grok-imagine-image-2-0\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"grok-imagine-image-2-0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-6:free": {
"id": "kimi-k2-6:free",
"name": "Kimi K2.6 (Free)",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/kimi-k2-6:free\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-6:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-6": {
"id": "kimi-k2-6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/kimi-k2-6\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-1-flash-lite": {
"id": "gemini-3-1-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gemini-3-1-flash-lite\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"hy3:free": {
"id": "hy3:free",
"name": "Hy3 (Free)",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 192000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/hy3:free\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"hy3:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7": {
"id": "claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/claude-opus-4-7\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/kimi-k3\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-large:free": {
"id": "mistral-large:free",
"name": "Mistral Large (Free)",
"description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
"family": "mistral-large",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-11",
"release_date": "2024-11-01",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/mistral-large:free\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"mistral-large:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-5": {
"id": "grok-4-5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/grok-4-5\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-5:free": {
"id": "mimo-v2-5:free",
"name": "MiMo-V2.5 (Free)",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/mimo-v2-5:free\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-5:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/claude-fable-5\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-5": {
"id": "gpt-5-5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gpt-5-5\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5-3-flash": {
"id": "glm-5-3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/glm-5-3-flash\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"glm-5-3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-super-120b-a12b": {
"id": "nemotron-3-super-120b-a12b",
"name": "Nemotron 3 Super 120B A12B",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/nemotron-3-super-120b-a12b\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-super-120b-a12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-4-mini": {
"id": "gpt-5-4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gpt-5-4-mini\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-super-120b-a12b:free": {
"id": "nemotron-3-super-120b-a12b:free",
"name": "Nemotron 3 Super 120B A12B (Free)",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/nemotron-3-super-120b-a12b:free\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-super-120b-a12b:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"whisper-large-v3-turbo": {
"id": "whisper-large-v3-turbo",
"name": "Whisper Large v3 Turbo",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "whisper",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2024-10-01",
"last_updated": "2024-10-01",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 448,
"output": 448
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/whisper-large-v3-turbo\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"whisper-large-v3-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-build-0-1": {
"id": "grok-build-0-1",
"name": "Grok Build 0.1",
"description": "Fast Grok coding model tuned for agentic engineering and iterative edits",
"family": "grok-build",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/grok-build-0-1\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"grok-build-0-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5-2": {
"id": "glm-5-2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/glm-5-2\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"glm-5-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4-7-flash:free": {
"id": "glm-4-7-flash:free",
"name": "GLM-4.7-Flash (Free)",
"description": "Budget GLM lane for fast coding help, routing, and everyday automation",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/glm-4-7-flash:free\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"glm-4-7-flash:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it": {
"id": "gemma-4-31b-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gemma-4-31b-it\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-5": {
"id": "mimo-v2-5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/mimo-v2-5\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-image-2": {
"id": "gpt-image-2",
"name": "GPT-Image-2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-image",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 272000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gpt-image-2\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gpt-image-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-5-pro": {
"id": "mimo-v2-5-pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/mimo-v2-5-pro\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-6": {
"id": "grok-4-6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/grok-4-6\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3-7-flash": {
"id": "step-3-7-flash",
"name": "Step 3.7 Flash",
"description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/step-3-7-flash\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"step-3-7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/claude-opus-4-8\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/deepseek-v4-pro\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2-5-flash-lite": {
"id": "gemini-2-5-flash-lite",
"name": "Gemini 2.5 Flash-Lite",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gemini-2-5-flash-lite\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2-5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gpt-oss-120b\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-8-max": {
"id": "qwen3-8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/qwen3-8-max\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-ultra-550b-a55b": {
"id": "nemotron-3-ultra-550b-a55b",
"name": "Nemotron 3 Ultra 550B A55B",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/nemotron-3-ultra-550b-a55b\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-ultra-550b-a55b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2-7": {
"id": "minimax-m2-7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/minimax-m2-7\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5": {
"id": "claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/claude-sonnet-5\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-1-flash-tts": {
"id": "gemini-3-1-flash-tts",
"name": "Gemini 3.1 Flash TTS Preview",
"description": "Low-latency speech generation with steerable prompts and expressive audio tags",
"family": "gemini-flash",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-15",
"last_updated": "2026-04-15",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/gemini-3-1-flash-tts\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-1-flash-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-nano-30b-a3b": {
"id": "nemotron-3-nano-30b-a3b",
"name": "Nemotron 3 Nano 30B A3B",
"description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-15",
"last_updated": "2025-12-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kenari/nemotron-3-nano-30b-a3b\", apiKey: processEnvironment[\"KENARI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://kenari.id/v1\")!,\n apiKey: processEnvironment[\"KENARI_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-nano-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"the-grid-ai": {
"id": "the-grid-ai",
"name": "The Grid AI",
"baseURL": "https://api.thegrid.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"THEGRID_API_KEY"
],
"doc": "https://thegrid.ai/docs",
"modelCount": 9,
"models": {
"agent-prime": {
"id": "agent-prime",
"name": "Agent Prime",
"description": "Reliable models for dependable agentic applications, multi-step tool use, and reasoning workflows. Any model that meets the contract spec can serve your request.",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-05-04",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 196608,
"input": 120000,
"output": 30000
},
"status": "beta",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"the-grid-ai/agent-prime\", apiKey: processEnvironment[\"THEGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.thegrid.ai/v1\")!,\n apiKey: processEnvironment[\"THEGRID_API_KEY\"]\n)\nlet session = provider.model(\"agent-prime\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"text-standard": {
"id": "text-standard",
"name": "Text Standard",
"description": "Price-optimized models with low-latency, high-throughput and shorter maximum outputs. Any model that meets the contract spec can serve your request.",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-26",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 120000,
"output": 16000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"the-grid-ai/text-standard\", apiKey: processEnvironment[\"THEGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.thegrid.ai/v1\")!,\n apiKey: processEnvironment[\"THEGRID_API_KEY\"]\n)\nlet session = provider.model(\"text-standard\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"agent-max": {
"id": "agent-max",
"name": "Agent Max",
"description": "Frontier models for autonomous research, deep multi-step tool chains, and complex long-horizon tasks. Any model that meets the contract spec can serve your request.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-05-04",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"status": "beta",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"the-grid-ai/agent-max\", apiKey: processEnvironment[\"THEGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.thegrid.ai/v1\")!,\n apiKey: processEnvironment[\"THEGRID_API_KEY\"]\n)\nlet session = provider.model(\"agent-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"text-max": {
"id": "text-max",
"name": "Text Max",
"description": "Frontier models for deep reasoning, long context, and complex workflows. Any model that meets the contract spec can serve your request.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-02-26",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"the-grid-ai/text-max\", apiKey: processEnvironment[\"THEGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.thegrid.ai/v1\")!,\n apiKey: processEnvironment[\"THEGRID_API_KEY\"]\n)\nlet session = provider.model(\"text-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"code-max": {
"id": "code-max",
"name": "Code Max",
"description": "Frontier models for complex research, architectural decisions, debugging, and multi-file development. Any model that meets the contract spec can serve your request.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-05-04",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"status": "beta",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"the-grid-ai/code-max\", apiKey: processEnvironment[\"THEGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.thegrid.ai/v1\")!,\n apiKey: processEnvironment[\"THEGRID_API_KEY\"]\n)\nlet session = provider.model(\"code-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"code-standard": {
"id": "code-standard",
"name": "Code Standard",
"description": "Price-optimized models for rapid autocomplete, linting, high-frequency suggestions, and batch edits. Any model that meets the contract spec can serve your request.",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-05-04",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 120000,
"output": 16000
},
"status": "beta",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"the-grid-ai/code-standard\", apiKey: processEnvironment[\"THEGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.thegrid.ai/v1\")!,\n apiKey: processEnvironment[\"THEGRID_API_KEY\"]\n)\nlet session = provider.model(\"code-standard\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"code-prime": {
"id": "code-prime",
"name": "Code Prime",
"description": "Reliable models for everyday software tasks, code completion, review, and standard debugging. Any model that meets the contract spec can serve your request.",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-05-04",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 196608,
"input": 120000,
"output": 30000
},
"status": "beta",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"the-grid-ai/code-prime\", apiKey: processEnvironment[\"THEGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.thegrid.ai/v1\")!,\n apiKey: processEnvironment[\"THEGRID_API_KEY\"]\n)\nlet session = provider.model(\"code-prime\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"text-prime": {
"id": "text-prime",
"name": "Text Prime",
"description": "Reliable models for everyday text generation, editing, and analysis across diverse workflows. Any model that meets the contract spec can serve your request.",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-26",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 196608,
"input": 120000,
"output": 30000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"the-grid-ai/text-prime\", apiKey: processEnvironment[\"THEGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.thegrid.ai/v1\")!,\n apiKey: processEnvironment[\"THEGRID_API_KEY\"]\n)\nlet session = provider.model(\"text-prime\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"agent-standard": {
"id": "agent-standard",
"name": "Agent Standard",
"description": "Price-optimized models for fast tool calls, simple agent loops, high-throughput automation, and orchestration. Any model that meets the contract spec can serve your request.",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-05-04",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 120000,
"output": 16000
},
"status": "beta",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"the-grid-ai/agent-standard\", apiKey: processEnvironment[\"THEGRID_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.thegrid.ai/v1\")!,\n apiKey: processEnvironment[\"THEGRID_API_KEY\"]\n)\nlet session = provider.model(\"agent-standard\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"google-vertex": {
"id": "google-vertex",
"name": "Vertex",
"baseURL": "",
"npm": "@ai-sdk/google-vertex",
"swiftDriver": "geminiNative",
"env": [
"GOOGLE_VERTEX_PROJECT",
"GOOGLE_VERTEX_LOCATION",
"GOOGLE_APPLICATION_CREDENTIALS"
],
"doc": "https://cloud.google.com/vertex-ai/generative-ai/docs/models",
"modelCount": 51,
"models": {
"gemini-2.5-flash-tts": {
"id": "gemini-2.5-flash-tts",
"name": "Gemini 2.5 Flash TTS",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "gemini-flash",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-09-30",
"last_updated": "2025-12-10",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0.5,
"output": 10
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-2.5-flash-tts\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro-preview-customtools": {
"id": "gemini-3.1-pro-preview-customtools",
"name": "Gemini 3.1 Pro Preview Custom Tools",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3.1-pro-preview-customtools\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3.1-pro-preview-customtools\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro-tts": {
"id": "gemini-2.5-pro-tts",
"name": "Gemini 2.5 Pro TTS",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "gemini-pro",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-09-30",
"last_updated": "2025-12-10",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 1,
"output": 20
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-2.5-pro-tts\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-2.5-pro-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4@20250514": {
"id": "claude-sonnet-4@20250514",
"name": "Claude Sonnet 4",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-sonnet-4@20250514\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-sonnet-4@20250514\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-image": {
"id": "gemini-2.5-flash-image",
"name": "Nano Banana",
"description": "Nano Banana image model for fast generation, edits, and character-consistent assets",
"family": "gemini-flash",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2025-08-26",
"last_updated": "2025-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 30
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-2.5-flash-image\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5@20251101": {
"id": "claude-opus-4-5@20251101",
"name": "Claude Opus 4.5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-01",
"last_updated": "2025-11-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-opus-4-5@20251101\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-opus-4-5@20251101\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-pro-image": {
"id": "gemini-3-pro-image",
"name": "Nano Banana Pro",
"description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 32768
},
"cost": {
"input": 2,
"output": 120,
"cache_read": 0.2
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3-pro-image\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3-pro-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro-preview": {
"id": "gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite": {
"id": "gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash-Lite",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 512,
"max": 24576
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01,
"input_audio": 0.3
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-6@default": {
"id": "claude-sonnet-4-6@default",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75,
"tiers": [
{
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-sonnet-4-6@default\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6@default\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.6-flash": {
"id": "gemini-3.6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"input_audio": 0.75
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3.6-flash\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5@default": {
"id": "claude-fable-5@default",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-fable-5@default\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-fable-5@default\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite": {
"id": "gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"input_audio": 0.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6@default": {
"id": "claude-opus-4-6@default",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-opus-4-6@default\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-opus-4-6@default\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4@20250514": {
"id": "claude-opus-4@20250514",
"name": "Claude Opus 4",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-opus-4@20250514\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-opus-4@20250514\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5@20251001": {
"id": "claude-haiku-4-5@20251001",
"name": "Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-haiku-4-5@20251001\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-haiku-4-5@20251001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5@default": {
"id": "claude-sonnet-5@default",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-sonnet-5@default\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-sonnet-5@default\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash": {
"id": "gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"input_audio": 1.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3.5-flash\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite-preview": {
"id": "gemini-3.1-flash-lite-preview",
"name": "Gemini 3.1 Flash Lite Preview",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"status": "deprecated",
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"input_audio": 0.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3.1-flash-lite-preview\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1@20250805": {
"id": "claude-opus-4-1@20250805",
"name": "Claude Opus 4.1",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-opus-4-1@20250805\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-opus-4-1@20250805\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-embedding-001": {
"id": "gemini-embedding-001",
"name": "Gemini Embedding 001",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "gemini",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2025-05",
"release_date": "2025-05-20",
"last_updated": "2025-05-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2048,
"output": 1
},
"cost": {
"input": 0.15,
"output": 0
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-embedding-001\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-embedding-001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-image": {
"id": "gemini-3.1-flash-image",
"name": "Nano Banana 2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 60,
"cache_read": 0.05
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3.1-flash-image\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash-lite": {
"id": "gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5-1@default": {
"id": "claude-fable-5-1@default",
"name": "Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-fable-5-1@default\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-fable-5-1@default\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7@default": {
"id": "claude-opus-4-7@default",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-opus-4-7@default\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-opus-4-7@default\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-flash-lite-latest": {
"id": "gemini-flash-lite-latest",
"name": "Gemini Flash-Lite Latest",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"input_audio": 0.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-flash-lite-latest\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-flash-lite-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5@default": {
"id": "claude-opus-5@default",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-opus-5@default\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-opus-5@default\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-flash-preview": {
"id": "gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"input_audio": 1
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3-flash-preview\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.8-flash": {
"id": "gemini-3.8-flash",
"name": "Gemini 3.8 Flash",
"description": "Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"input_audio": 0.75
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3.8-flash\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.7-flash": {
"id": "gemini-3.7-flash",
"name": "Gemini 3.7 Flash",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"input_audio": 0.75
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-3.7-flash\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-2.5-pro\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-flash-latest": {
"id": "gemini-flash-latest",
"name": "Gemini Flash Latest",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"input_audio": 1.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-flash-latest\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-flash-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"input_audio": 1
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/gemini-2.5-flash\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8@default": {
"id": "claude-opus-4-8@default",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 10,
"output": 37.5,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-opus-4-8@default\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-opus-4-8@default\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5@20250929": {
"id": "claude-sonnet-4-5@20250929",
"name": "Claude Sonnet 4.5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"provider": {
"npm": "@ai-sdk/google-vertex/anthropic"
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/claude-sonnet-4-5@20250929\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5@20250929\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b-instruct-2507-maas": {
"id": "qwen/qwen3-235b-a22b-instruct-2507-maas",
"name": "Qwen3 235B A22B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-13",
"last_updated": "2025-08-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 0.22,
"output": 0.88
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/qwen/qwen3-235b-a22b-instruct-2507-maas\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b-instruct-2507-maas\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/deepseek-v3.1-maas": {
"id": "deepseek-ai/deepseek-v3.1-maas",
"name": "DeepSeek V3.1",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-28",
"last_updated": "2025-08-28",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 32768
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 0.6,
"output": 1.7,
"cache_read": 0.06
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/deepseek-ai/deepseek-v3.1-maas\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"deepseek-ai/deepseek-v3.1-maas\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/deepseek-v3.2-maas": {
"id": "deepseek-ai/deepseek-v3.2-maas",
"name": "DeepSeek V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-17",
"last_updated": "2026-04-04",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 65536
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 0.56,
"output": 1.68,
"cache_read": 0.056
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/deepseek-ai/deepseek-v3.2-maas\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"deepseek-ai/deepseek-v3.2-maas\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-5-maas": {
"id": "zai-org/glm-5-maas",
"name": "GLM-5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-11",
"last_updated": "2026-02-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.1
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/zai-org/glm-5-maas\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"zai-org/glm-5-maas\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.7-maas": {
"id": "zai-org/glm-4.7-maas",
"name": "GLM-4.7",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-06",
"last_updated": "2026-01-06",
"modalities": {
"input": [
"text",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.06
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/zai-org/glm-4.7-maas\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"zai-org/glm-4.7-maas\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-4-maverick-17b-128e-instruct-maas": {
"id": "meta/llama-4-maverick-17b-128e-instruct-maas",
"name": "Llama 4 Maverick 17B 128E Instruct",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-29",
"last_updated": "2025-04-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 8192
},
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 0.35,
"output": 1.15
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/meta/llama-4-maverick-17b-128e-instruct-maas\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"meta/llama-4-maverick-17b-128e-instruct-maas\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-3.3-70b-instruct-maas": {
"id": "meta/llama-3.3-70b-instruct-maas",
"name": "Llama 3.3 70B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2025-04-29",
"last_updated": "2025-04-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 0.72,
"output": 0.72
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/meta/llama-3.3-70b-instruct-maas\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"meta/llama-3.3-70b-instruct-maas\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b-maas": {
"id": "openai/gpt-oss-120b-maas",
"name": "GPT OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.09,
"output": 0.36
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/openai/gpt-oss-120b-maas\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b-maas\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b-maas": {
"id": "openai/gpt-oss-20b-maas",
"name": "GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"status": "deprecated",
"cost": {
"input": 0.07,
"output": 0.25,
"cache_read": 0.007
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/openai/gpt-oss-20b-maas\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b-maas\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-thinking-maas": {
"id": "moonshotai/kimi-k2-thinking-maas",
"name": "Kimi K2 Thinking",
"description": "Kimi reasoning model for long-horizon research, planning, and tool use",
"family": "kimi-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 0.6,
"output": 2.5,
"cache_read": 0.06
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/moonshotai/kimi-k2-thinking-maas\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-thinking-maas\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.20-reasoning": {
"id": "xai/grok-4.20-reasoning",
"name": "Grok 4.20 (Reasoning)",
"description": "Reasoning Grok for document-heavy analysis and long-horizon tool use",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/xai/grok-4.20-reasoning\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"xai/grok-4.20-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.3": {
"id": "xai/grok-4.3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 30000
},
"status": "beta",
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/xai/grok-4.3\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"xai/grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.20-non-reasoning": {
"id": "xai/grok-4.20-non-reasoning",
"name": "Grok 4.20 (Non-Reasoning)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/xai/grok-4.20-non-reasoning\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"xai/grok-4.20-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.1-fast-non-reasoning": {
"id": "xai/grok-4.1-fast-non-reasoning",
"name": "Grok 4.1 Fast",
"description": "Fast Grok model for responsive chat, tool-assisted work, and low-latency responses",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 30000
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/xai/grok-4.1-fast-non-reasoning\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"xai/grok-4.1-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.1-fast-reasoning": {
"id": "xai/grok-4.1-fast-reasoning",
"name": "Grok 4.1 Fast (Reasoning)",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 30000
},
"status": "deprecated",
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/xai/grok-4.1-fast-reasoning\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"xai/grok-4.1-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.6": {
"id": "xai/grok-4.6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 524288,
"output": 500000
},
"status": "beta",
"provider": {
"npm": "@ai-sdk/openai-compatible",
"api": "https://${GOOGLE_VERTEX_ENDPOINT}/v1/projects/${GOOGLE_VERTEX_PROJECT}/locations/${GOOGLE_VERTEX_LOCATION}/endpoints/openapi"
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5,
"tiers": [
{
"input": 4,
"output": 12,
"cache_read": 1,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 12,
"cache_read": 1
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google-vertex/xai/grok-4.6\", apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_VERTEX_PROJECT\"]\n)\nlet session = provider.model(\"xai/grok-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"stepfun-ai": {
"id": "stepfun-ai",
"name": "StepFun (Global)",
"baseURL": "https://api.stepfun.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"STEPFUN_API_KEY"
],
"doc": "https://platform.stepfun.ai/docs/en/overview/concept",
"modelCount": 8,
"models": {
"step-1-32k": {
"id": "step-1-32k",
"name": "Step 1 (32K)",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2025-01-01",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 2.05,
"output": 9.59,
"cache_read": 0.41
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun-ai/step-1-32k\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.ai/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-1-32k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepaudio-2.5-tts": {
"id": "stepaudio-2.5-tts",
"name": "StepAudio 2.5 TTS",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "step",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-04-16",
"last_updated": "2026-07-02",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun-ai/stepaudio-2.5-tts\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.ai/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"stepaudio-2.5-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3.5-flash-2603": {
"id": "step-3.5-flash-2603",
"name": "Step 3.5 Flash 2603",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun-ai/step-3.5-flash-2603\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.ai/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-3.5-flash-2603\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepaudio-2.5-asr": {
"id": "stepaudio-2.5-asr",
"name": "StepAudio 2.5 ASR",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "step",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-04-24",
"last_updated": "2026-07-02",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun-ai/stepaudio-2.5-asr\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.ai/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"stepaudio-2.5-asr\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-2-16k": {
"id": "step-2-16k",
"name": "Step 2 (16K)",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2025-01-01",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 5.21,
"output": 16.44,
"cache_read": 1.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun-ai/step-2-16k\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.ai/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-2-16k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3.5-flash": {
"id": "step-3.5-flash",
"name": "Step 3.5 Flash",
"description": "StepFun flash lane for quick multimodal reasoning and coding assistance",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-29",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun-ai/step-3.5-flash\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.ai/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-tts-2": {
"id": "step-tts-2",
"name": "Step TTS 2",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "step",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-03-01",
"last_updated": "2026-07-02",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun-ai/step-tts-2\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.ai/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-tts-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3.7-flash": {
"id": "step-3.7-flash",
"name": "Step 3.7 Flash",
"description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-06-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"cost": {
"input": 0.185,
"output": 1.11,
"cache_read": 0.037
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun-ai/step-3.7-flash\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.ai/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"pendra": {
"id": "pendra",
"name": "Pendra",
"baseURL": "https://api.pendra.ai/api/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"PENDRA_API_KEY"
],
"doc": "https://pendra.ai/docs/integrations/opencode",
"modelCount": 6,
"models": {
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pendra/deepseek-v4-flash\", apiKey: processEnvironment[\"PENDRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pendra.ai/api/v1\")!,\n apiKey: processEnvironment[\"PENDRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss:120b": {
"id": "gpt-oss:120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pendra/gpt-oss:120b\", apiKey: processEnvironment[\"PENDRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pendra.ai/api/v1\")!,\n apiKey: processEnvironment[\"PENDRA_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss:120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder:30b": {
"id": "qwen3-coder:30b",
"name": "Qwen3-Coder 30B-A3B Instruct",
"description": "Smaller Qwen coder for efficient local agents and repo-level fixes",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pendra/qwen3-coder:30b\", apiKey: processEnvironment[\"PENDRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pendra.ai/api/v1\")!,\n apiKey: processEnvironment[\"PENDRA_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder:30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6:27b": {
"id": "qwen3.6:27b",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pendra/qwen3.6:27b\", apiKey: processEnvironment[\"PENDRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pendra.ai/api/v1\")!,\n apiKey: processEnvironment[\"PENDRA_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6:27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama3.3:70b": {
"id": "llama3.3:70b",
"name": "Llama-3.3-70B-Instruct",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pendra/llama3.3:70b\", apiKey: processEnvironment[\"PENDRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pendra.ai/api/v1\")!,\n apiKey: processEnvironment[\"PENDRA_API_KEY\"]\n)\nlet session = provider.model(\"llama3.3:70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.7-flash": {
"id": "glm-4.7-flash",
"name": "GLM-4.7-Flash",
"description": "Budget GLM lane for fast coding help, routing, and everyday automation",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pendra/glm-4.7-flash\", apiKey: processEnvironment[\"PENDRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pendra.ai/api/v1\")!,\n apiKey: processEnvironment[\"PENDRA_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"above": {
"id": "above",
"name": "above.dev",
"baseURL": "https://api.above.dev/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ABOVE_API_KEY"
],
"doc": "https://above.dev/docs",
"modelCount": 8,
"models": {
"deepseek-v4-flash-vision-exp": {
"id": "deepseek-v4-flash-vision-exp",
"name": "DeepSeek V4 Flash Vision (Exp)",
"description": "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-21",
"last_updated": "2026-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.242,
"output": 0.726,
"reasoning": 0.726,
"cache_read": 0.0077
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"above/deepseek-v4-flash-vision-exp\", apiKey: processEnvironment[\"ABOVE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.above.dev/v1\")!,\n apiKey: processEnvironment[\"ABOVE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-vision-exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM 5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.54,
"output": 4.84,
"cache_read": 0.154
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"above/glm-5.2\", apiKey: processEnvironment[\"ABOVE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.above.dev/v1\")!,\n apiKey: processEnvironment[\"ABOVE_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.165,
"output": 0.66,
"reasoning": 0.66,
"cache_read": 0.0033
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"above/deepseek-v4-flash\", apiKey: processEnvironment[\"ABOVE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.above.dev/v1\")!,\n apiKey: processEnvironment[\"ABOVE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-fast": {
"id": "glm-5.2-fast",
"name": "GLM 5.2 Fast",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2.31,
"output": 7.26,
"cache_read": 0.231
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"above/glm-5.2-fast\", apiKey: processEnvironment[\"ABOVE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.above.dev/v1\")!,\n apiKey: processEnvironment[\"ABOVE_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM 5.3 Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.165,
"output": 0.55,
"cache_read": 0.0319
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"above/glm-5.3-flash\", apiKey: processEnvironment[\"ABOVE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.above.dev/v1\")!,\n apiKey: processEnvironment[\"ABOVE_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max": {
"id": "qwen3.8-max",
"name": "Qwen 3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 2.2,
"output": 6.6,
"cache_read": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"above/qwen3.8-max\", apiKey: processEnvironment[\"ABOVE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.above.dev/v1\")!,\n apiKey: processEnvironment[\"ABOVE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.726,
"output": 2.178,
"reasoning": 2.178,
"cache_read": 0.0242
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"above/deepseek-v4-pro\", apiKey: processEnvironment[\"ABOVE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.above.dev/v1\")!,\n apiKey: processEnvironment[\"ABOVE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-pro": {
"id": "mimo-v2.5-pro",
"name": "MiMo V2.5 Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.5077,
"output": 1.0154,
"cache_read": 0.0042
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"above/mimo-v2.5-pro\", apiKey: processEnvironment[\"ABOVE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.above.dev/v1\")!,\n apiKey: processEnvironment[\"ABOVE_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"scaleway": {
"id": "scaleway",
"name": "Scaleway",
"baseURL": "https://api.scaleway.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"SCALEWAY_API_KEY"
],
"doc": "https://www.scaleway.com/en/docs/generative-apis/",
"modelCount": 15,
"models": {
"gemma-4-26b-a4b-it": {
"id": "gemma-4-26b-a4b-it",
"name": "Gemma 4 26B A4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-01",
"last_updated": "2026-05-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"status": "beta",
"cost": {
"input": 0.25,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"status": "beta",
"cost": {
"input": 0.468,
"output": 0.936,
"reasoning": 0.936,
"cache_read": 0.0936
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 1.8,
"output": 5.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/glm-5.2\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-30b-a3b-instruct": {
"id": "qwen3-coder-30b-a3b-instruct",
"name": "Qwen3-Coder 30B-A3B Instruct",
"description": "Smaller Qwen coder for efficient local agents and repo-level fixes",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-397b-a17b": {
"id": "qwen3.5-397b-a17b",
"name": "Qwen3.5 397B A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-embedding-8b": {
"id": "qwen3-embedding-8b",
"name": "Qwen3 Embedding 8B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-06-05",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 4096
},
"cost": {
"input": 0.1,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/qwen3-embedding-8b\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-embedding-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"whisper-large-v3": {
"id": "whisper-large-v3",
"name": "Whisper Large v3",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "whisper",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2023-09",
"release_date": "2023-09-01",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 8192
},
"cost": {
"input": 0.003,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/whisper-large-v3\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"whisper-large-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-35b-a3b": {
"id": "qwen3.6-35b-a3b",
"name": "Qwen3.6 35B A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-05-01",
"last_updated": "2026-05-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"status": "beta",
"cost": {
"input": 0.25,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/qwen3.6-35b-a3b\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bge-multilingual-gemma2": {
"id": "bge-multilingual-gemma2",
"name": "BGE Multilingual Gemma2",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-07-26",
"last_updated": "2025-06-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8191,
"output": 3072
},
"cost": {
"input": 0.1,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/bge-multilingual-gemma2\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bge-multilingual-gemma2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-medium-3.5-128b": {
"id": "mistral-medium-3.5-128b",
"name": "Mistral Medium 3.5 128B",
"description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
"family": "mistral-medium",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-29",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 1.5,
"output": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/mistral-medium-3.5-128b\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral-medium-3.5-128b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-3.2-24b-instruct-2506": {
"id": "mistral-small-3.2-24b-instruct-2506",
"name": "Mistral Small 3.2 24B Instruct (2506)",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-06-20",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/mistral-small-3.2-24b-instruct-2506\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small-3.2-24b-instruct-2506\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-instruct-2507": {
"id": "qwen3-235b-a22b-instruct-2507",
"name": "Qwen3 235B A22B Instruct 2507",
"description": "Large open Qwen MoE for multilingual reasoning, coding, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-01",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 260000,
"output": 16384
},
"cost": {
"input": 0.75,
"output": 2.25,
"reasoning": 8.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/qwen3-235b-a22b-instruct-2507\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT-OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2024-01-01",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/gpt-oss-120b\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"pixtral-12b-2409": {
"id": "pixtral-12b-2409",
"name": "Pixtral 12B 2409",
"description": "Mistral vision-language model for image understanding and multimodal chat",
"family": "pixtral",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2024-09-25",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.2,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/pixtral-12b-2409\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"pixtral-12b-2409\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-70b-instruct": {
"id": "llama-3.3-70b-instruct",
"name": "Llama-3.3-70B-Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 100000,
"output": 16384
},
"cost": {
"input": 0.9,
"output": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"scaleway/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"SCALEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.scaleway.ai/v1\")!,\n apiKey: processEnvironment[\"SCALEWAY_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"alibaba-cn": {
"id": "alibaba-cn",
"name": "Alibaba (China)",
"baseURL": "https://dashscope.aliyuncs.com/compatible-mode/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"DASHSCOPE_API_KEY"
],
"doc": "https://www.alibabacloud.com/help/en/model-studio/models",
"modelCount": 87,
"models": {
"qwen3.7-max": {
"id": "qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 262144
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.5,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3.7-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-72b-instruct": {
"id": "qwen2-5-72b-instruct",
"name": "Qwen2.5 72B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.574,
"output": 1.721
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen2-5-72b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1-distill-qwen-7b": {
"id": "deepseek-r1-distill-qwen-7b",
"name": "DeepSeek R1 Distill Qwen 7B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0.072,
"output": 0.144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-r1-distill-qwen-7b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1-distill-qwen-7b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-next-80b-a3b-thinking": {
"id": "qwen3-next-80b-a3b-thinking",
"name": "Qwen3-Next 80B-A3B (Thinking)",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.144,
"output": 1.434
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-next-80b-a3b-thinking\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-next-80b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3": {
"id": "deepseek-v3",
"name": "DeepSeek V3",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-12-01",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 8192
},
"cost": {
"input": 0.287,
"output": 1.147
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-v3\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-omni-7b": {
"id": "qwen2-5-omni-7b",
"name": "Qwen2.5-Omni 7B",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-12",
"last_updated": "2024-12",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text",
"audio"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 2048
},
"cost": {
"input": 0.087,
"output": 0.345,
"input_audio": 5.448
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen2-5-omni-7b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-omni-7b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-mt-turbo": {
"id": "qwen-mt-turbo",
"name": "Qwen-MT Turbo",
"description": "Translation model for multilingual conversion, localization, and cross-language workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-01",
"last_updated": "2025-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"output": 8192
},
"cost": {
"input": 0.101,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-mt-turbo\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-mt-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3-1": {
"id": "deepseek-v3-1",
"name": "DeepSeek V3.1",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0.574,
"output": 1.721
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-v3-1\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-deep-research": {
"id": "qwen-deep-research",
"name": "Qwen Deep Research",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01",
"last_updated": "2024-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 7.742,
"output": 23.367
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-deep-research\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-deep-research\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-vl-max": {
"id": "qwen-vl-max",
"name": "Qwen-VL Max",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-04-08",
"last_updated": "2025-08-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.23,
"output": 0.574
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-vl-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-vl-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-next-80b-a3b-instruct": {
"id": "qwen3-next-80b-a3b-instruct",
"name": "Qwen3-Next 80B-A3B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.144,
"output": 0.574
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-flash": {
"id": "qwen3-coder-flash",
"name": "Qwen3 Coder Flash",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.144,
"output": 0.574
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-coder-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-14b": {
"id": "qwen3-14b",
"name": "Qwen3 14B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 38912
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.144,
"output": 0.574,
"reasoning": 1.434
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-14b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-14b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-max": {
"id": "qwen-max",
"name": "Qwen Max",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-04-03",
"last_updated": "2025-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.345,
"output": 1.377
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-plus": {
"id": "qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 81920
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"cache_write": 0.625,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3.6-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot-kimi-k2-instruct": {
"id": "moonshot-kimi-k2-instruct",
"name": "Moonshot Kimi K2 Instruct",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.574,
"output": 2.294
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/moonshot-kimi-k2-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"moonshot-kimi-k2-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-vl-plus": {
"id": "qwen-vl-plus",
"name": "Qwen-VL Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01-25",
"last_updated": "2025-08-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.115,
"output": 0.287
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-vl-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-vl-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-omni-turbo-realtime": {
"id": "qwen-omni-turbo-realtime",
"name": "Qwen-Omni Turbo Realtime",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-05-08",
"last_updated": "2025-05-08",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 2048
},
"cost": {
"input": 0.23,
"output": 0.918,
"input_audio": 3.584,
"output_audio": 7.168
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-omni-turbo-realtime\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-omni-turbo-realtime\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Moonshot Kimi K2.6",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 81920
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.929,
"output": 3.858
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/kimi-k2.6\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-flash": {
"id": "qwen-flash",
"name": "Qwen Flash",
"description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 81920
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.022,
"output": 0.216
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.1,
"output": 3.851,
"cache_read": 0.275,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/glm-5.2\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-turbo": {
"id": "qwen-turbo",
"name": "Qwen Turbo",
"description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 38912
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-11-01",
"last_updated": "2025-07-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 16384
},
"cost": {
"input": 0.044,
"output": 0.087,
"reasoning": 0.431
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-turbo\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-30b-a3b": {
"id": "qwen3-vl-30b-a3b",
"name": "Qwen3-VL 30B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.108,
"output": 0.431,
"reasoning": 1.076
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-vl-30b-a3b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-vl-ocr": {
"id": "qwen-vl-ocr",
"name": "Qwen-VL OCR",
"description": "OCR model for extracting structured text from documents and screenshots",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-10-28",
"last_updated": "2025-04-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 34096,
"output": 4096
},
"cost": {
"input": 0.717,
"output": 0.717
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-vl-ocr\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-vl-ocr\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tongyi-intent-detect-v3": {
"id": "tongyi-intent-detect-v3",
"name": "Tongyi Intent Detect V3",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"family": "yi",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01",
"last_updated": "2024-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1024
},
"cost": {
"input": 0.058,
"output": 0.144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/tongyi-intent-detect-v3\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"tongyi-intent-detect-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-v4-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-thinking": {
"id": "kimi-k2-thinking",
"name": "Moonshot Kimi K2 Thinking",
"description": "Kimi reasoning model for long-horizon research, planning, and tool use",
"family": "kimi-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-11-06",
"last_updated": "2025-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.574,
"output": 2.294
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/kimi-k2-thinking\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-32b": {
"id": "qwen3-32b",
"name": "Qwen3 32B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 38912
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.287,
"output": 1.147,
"reasoning": 2.868
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-32b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwq-plus": {
"id": "qwq-plus",
"name": "QwQ Plus",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-03-05",
"last_updated": "2025-03-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.23,
"output": 0.574
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwq-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwq-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-flash": {
"id": "qwen3.5-flash",
"name": "Qwen3.5 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 81920
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.172,
"output": 1.72,
"reasoning": 1.72
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3.5-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-plus": {
"id": "qwen3-vl-plus",
"name": "Qwen3-VL Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 81920
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.143353,
"output": 1.433525,
"reasoning": 4.300576
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-vl-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3-2-exp": {
"id": "deepseek-v3-2-exp",
"name": "DeepSeek V3.2 Exp",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0.287,
"output": 0.431
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-v3-2-exp\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3-2-exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-7b-instruct": {
"id": "qwen2-5-7b-instruct",
"name": "Qwen2.5 7B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.072,
"output": 0.144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen2-5-7b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-30b-a3b-instruct": {
"id": "qwen3-coder-30b-a3b-instruct",
"name": "Qwen3-Coder 30B-A3B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.216,
"output": 0.861,
"tiers": [
{
"input": 0.323,
"output": 1.291,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.538,
"output": 2.151,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-math-7b-instruct": {
"id": "qwen2-5-math-7b-instruct",
"name": "Qwen2.5-Math 7B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 4096,
"output": 3072
},
"cost": {
"input": 0.144,
"output": 0.287
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen2-5-math-7b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-math-7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-397b-a17b": {
"id": "qwen3.5-397b-a17b",
"name": "Qwen3.5 397B-A17B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 81920
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.172,
"output": 1.032,
"reasoning": 1.032,
"tiers": [
{
"input": 0.43,
"output": 2.58,
"reasoning": 2.58,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwq-32b": {
"id": "qwq-32b",
"name": "QwQ 32B",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-12",
"last_updated": "2024-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.287,
"output": 0.861
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwq-32b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwq-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1": {
"id": "deepseek-r1",
"name": "DeepSeek R1",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.574,
"output": 2.294
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-r1\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-asr-flash": {
"id": "qwen3-asr-flash",
"name": "Qwen3-ASR Flash",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2024-04",
"release_date": "2025-09-08",
"last_updated": "2025-09-08",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 53248,
"output": 4096
},
"cost": {
"input": 0.032,
"output": 0.032
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-asr-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-asr-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-omni-flash": {
"id": "qwen3-omni-flash",
"name": "Qwen3-Omni Flash",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-09-15",
"last_updated": "2025-09-15",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 16384
},
"cost": {
"input": 0.058,
"output": 0.23,
"input_audio": 3.584,
"output_audio": 7.168
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-omni-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-omni-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1-distill-qwen-1-5b": {
"id": "deepseek-r1-distill-qwen-1-5b",
"name": "DeepSeek R1 Distill Qwen 1.5B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-r1-distill-qwen-1-5b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1-distill-qwen-1-5b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-flash": {
"id": "qwen3.7-flash",
"name": "Qwen3.7 Flash",
"description": "Lightweight multimodal Qwen model for high-throughput text, image, and video tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 262144
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 991000,
"output": 65536
},
"cost": {
"input": 0.02962,
"output": 0.1185,
"cache_read": 0.002962,
"cache_write": 0.03703,
"tiers": [
{
"input": 0.08887,
"output": 0.35549,
"cache_read": 0.008887,
"cache_write": 0.11109,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.17774,
"output": 0.71098,
"cache_read": 0.017774,
"cache_write": 0.22218,
"tier": {
"type": "context",
"size": 256000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3.7-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-vl-72b-instruct": {
"id": "qwen2-5-vl-72b-instruct",
"name": "Qwen2.5-VL 72B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 2.294,
"output": 6.881
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen2-5-vl-72b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-vl-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1-0528": {
"id": "deepseek-r1-0528",
"name": "DeepSeek R1 0528",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-05-28",
"last_updated": "2025-05-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.574,
"output": 2.294
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-r1-0528\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1-0528\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1-distill-qwen-32b": {
"id": "deepseek-r1-distill-qwen-32b",
"name": "DeepSeek R1 Distill Qwen 32B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0.287,
"output": 0.861
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-r1-distill-qwen-32b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1-distill-qwen-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-max": {
"id": "qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.861,
"output": 3.441
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-coder-32b-instruct": {
"id": "qwen2-5-coder-32b-instruct",
"name": "Qwen2.5-Coder 32B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-11",
"last_updated": "2024-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.287,
"output": 0.861
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen2-5-coder-32b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-coder-32b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1-distill-llama-70b": {
"id": "deepseek-r1-distill-llama-70b",
"name": "DeepSeek R1 Distill Llama 70B",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0.287,
"output": 0.861
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-r1-distill-llama-70b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1-distill-llama-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-math-72b-instruct": {
"id": "qwen2-5-math-72b-instruct",
"name": "Qwen2.5-Math 72B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 4096,
"output": 3072
},
"cost": {
"input": 0.574,
"output": 1.721
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen2-5-math-72b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-math-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-plus": {
"id": "qwen-plus",
"name": "Qwen Plus",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 81920
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01-25",
"last_updated": "2025-09-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.115,
"output": 0.287,
"reasoning": 1.147,
"cache_read": 0.012,
"cache_write": 0.144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.5": {
"id": "MiniMax-M2.5",
"name": "MiniMax-M2.5",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/MiniMax-M2.5\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-omni-flash-realtime": {
"id": "qwen3-omni-flash-realtime",
"name": "Qwen3-Omni Flash Realtime",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-09-15",
"last_updated": "2025-09-15",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 16384
},
"cost": {
"input": 0.23,
"output": 0.918,
"input_audio": 3.584,
"output_audio": 7.168
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-omni-flash-realtime\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-omni-flash-realtime\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-vl-7b-instruct": {
"id": "qwen2-5-vl-7b-instruct",
"name": "Qwen2.5-VL 7B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.287,
"output": 0.717
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen2-5-vl-7b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-vl-7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-flash": {
"id": "qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 131072
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.1875,
"output": 1.125,
"cache_write": 0.234375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3.6-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-flash": {
"id": "qwen3.8-flash",
"name": "Qwen3.8 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens",
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.11875,
"output": 0.40073,
"cache_read": 0.01187,
"cache_write": 0.14844
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3.8-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-14b-instruct": {
"id": "qwen2-5-14b-instruct",
"name": "Qwen2.5 14B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.144,
"output": 0.431
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen2-5-14b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-14b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-math-turbo": {
"id": "qwen-math-turbo",
"name": "Qwen Math Turbo",
"description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09-19",
"last_updated": "2024-09-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 4096,
"output": 3072
},
"cost": {
"input": 0.287,
"output": 0.861
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-math-turbo\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-math-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-plus-character": {
"id": "qwen-plus-character",
"name": "Qwen Plus Character",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01",
"last_updated": "2024-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 4096
},
"cost": {
"input": 0.115,
"output": 0.287
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-plus-character\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-plus-character\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-max-preview": {
"id": "qwen3.6-max-preview",
"name": "Qwen3.6 Max Preview",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 131072
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-20",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 245800,
"output": 65536
},
"cost": {
"input": 1.32,
"output": 7.9,
"cache_read": 0.132
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3.6-max-preview\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5": {
"id": "glm-5",
"name": "GLM-5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 32768
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-11",
"last_updated": "2026-02-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 202752,
"output": 16384
},
"cost": {
"input": 0.573,
"output": 2.58,
"tiers": [
{
"input": 0.86,
"output": 3.154,
"tier": {
"type": "context",
"size": 32000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/glm-5\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max": {
"id": "qwen3.8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens",
"min": 0,
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.77744,
"output": 5.33231,
"cache_read": 0.22218,
"cache_write": 2.22179
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3.8-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "Moonshot Kimi K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 81920
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.574,
"output": 2.411
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/kimi-k2.5\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-8b": {
"id": "qwen3-8b",
"name": "Qwen3 8B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 38912
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.072,
"output": 0.287,
"reasoning": 0.717
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-8b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 131072
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-14",
"last_updated": "2026-04-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 128000
},
"cost": {
"input": 0.825,
"output": 3.301,
"cache_read": 0.17,
"tiers": [
{
"input": 1.1,
"output": 3.851,
"tier": {
"type": "context",
"size": 32000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/glm-5.1\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b": {
"id": "qwen3-235b-a22b",
"name": "Qwen3 235B-A22B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 38912
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.287,
"output": 1.147,
"reasoning": 2.868
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-235b-a22b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-plus": {
"id": "qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 262144
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"cache_write": 0.625,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3.7-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-omni-turbo": {
"id": "qwen-omni-turbo",
"name": "Qwen-Omni Turbo",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-01-19",
"last_updated": "2025-03-26",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 2048
},
"cost": {
"input": 0.058,
"output": 0.23,
"input_audio": 3.584,
"output_audio": 7.168
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-omni-turbo\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-omni-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.003625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-v4-pro\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-480b-a35b-instruct": {
"id": "qwen3-coder-480b-a35b-instruct",
"name": "Qwen3-Coder 480B-A35B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.861,
"output": 3.441,
"tiers": [
{
"input": 1.291,
"output": 5.161,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 2.151,
"output": 8.602,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-coder-480b-a35b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-480b-a35b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-32b-instruct": {
"id": "qwen2-5-32b-instruct",
"name": "Qwen2.5 32B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.287,
"output": 0.861
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen2-5-32b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-32b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-coder-7b-instruct": {
"id": "qwen2-5-coder-7b-instruct",
"name": "Qwen2.5-Coder 7B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-11",
"last_updated": "2024-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.144,
"output": 0.287
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen2-5-coder-7b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-coder-7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-long": {
"id": "qwen-long",
"name": "Qwen Long",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-01-25",
"last_updated": "2025-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 10000000,
"output": 8192
},
"cost": {
"input": 0.072,
"output": 0.287
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-long\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-long\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1-distill-llama-8b": {
"id": "deepseek-r1-distill-llama-8b",
"name": "DeepSeek R1 Distill Llama 8B",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-r1-distill-llama-8b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1-distill-llama-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1-distill-qwen-14b": {
"id": "deepseek-r1-distill-qwen-14b",
"name": "DeepSeek R1 Distill Qwen 14B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0.144,
"output": 0.431
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/deepseek-r1-distill-qwen-14b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1-distill-qwen-14b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-235b-a22b": {
"id": "qwen3-vl-235b-a22b",
"name": "Qwen3-VL 235B-A22B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.286705,
"output": 1.14682,
"reasoning": 2.867051
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-vl-235b-a22b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-235b-a22b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-plus": {
"id": "qwen3.5-plus",
"name": "Qwen3.5 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"max": 81920
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.573,
"output": 3.44,
"reasoning": 3.44
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3.5-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-math-plus": {
"id": "qwen-math-plus",
"name": "Qwen Math Plus",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-08-16",
"last_updated": "2024-09-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 4096,
"output": 3072
},
"cost": {
"input": 0.574,
"output": 1.721
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-math-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-math-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-doc-turbo": {
"id": "qwen-doc-turbo",
"name": "Qwen Doc Turbo",
"description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01",
"last_updated": "2024-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.087,
"output": 0.144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-doc-turbo\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-doc-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-mt-plus": {
"id": "qwen-mt-plus",
"name": "Qwen-MT Plus",
"description": "Translation model for multilingual conversion, localization, and cross-language workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-01",
"last_updated": "2025-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"output": 8192
},
"cost": {
"input": 0.259,
"output": 0.775
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen-mt-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-mt-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qvq-max": {
"id": "qvq-max",
"name": "QVQ Max",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "qvq",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-03-25",
"last_updated": "2025-03-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 1.147,
"output": 4.588
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qvq-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qvq-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax/MiniMax-M2.7": {
"id": "MiniMax/MiniMax-M2.7",
"name": "MiniMax-M2.7",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/MiniMax/MiniMax-M2.7\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax/MiniMax-M2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"siliconflow/deepseek-r1-0528": {
"id": "siliconflow/deepseek-r1-0528",
"name": "siliconflow/deepseek-r1-0528",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-05-28",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 163840,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 2.18
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/siliconflow/deepseek-r1-0528\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"siliconflow/deepseek-r1-0528\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"siliconflow/deepseek-v3.2": {
"id": "siliconflow/deepseek-v3.2",
"name": "siliconflow/deepseek-v3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-03",
"last_updated": "2025-12-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 163840,
"output": 65536
},
"cost": {
"input": 0.27,
"output": 0.42
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/siliconflow/deepseek-v3.2\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"siliconflow/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"siliconflow/deepseek-v3.1-terminus": {
"id": "siliconflow/deepseek-v3.1-terminus",
"name": "siliconflow/deepseek-v3.1-terminus",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-29",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 163840,
"output": 65536
},
"cost": {
"input": 0.27,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/siliconflow/deepseek-v3.1-terminus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"siliconflow/deepseek-v3.1-terminus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"siliconflow/deepseek-v3-0324": {
"id": "siliconflow/deepseek-v3-0324",
"name": "siliconflow/deepseek-v3-0324",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-12-26",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 0.25,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/siliconflow/deepseek-v3-0324\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"siliconflow/deepseek-v3-0324\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi/kimi-k2.5": {
"id": "kimi/kimi-k2.5",
"name": "kimi/kimi-k2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/kimi/kimi-k2.5\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"kimi/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-plus": {
"id": "qwen3-coder-plus",
"name": "Qwen3 Coder Plus",
"description": "Hosted Qwen coder for software agents, repo edits, and long-context code",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba-cn/qwen3-coder-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"poe": {
"id": "poe",
"name": "Poe",
"baseURL": "https://api.poe.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"POE_API_KEY"
],
"doc": "https://creator.poe.com/docs/external-applications/openai-compatible-api",
"modelCount": 137,
"models": {
"poetools/claude-code": {
"id": "poetools/claude-code",
"name": "claude-code",
"description": "Claude model for careful reasoning, writing, coding, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-11-27",
"last_updated": "2025-11-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/poetools/claude-code\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"poetools/claude-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"elevenlabs/elevenlabs-v2.5-turbo": {
"id": "elevenlabs/elevenlabs-v2.5-turbo",
"name": "ElevenLabs-v2.5-Turbo",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "elevenlabs",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-10-28",
"last_updated": "2024-10-28",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/elevenlabs/elevenlabs-v2.5-turbo\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"elevenlabs/elevenlabs-v2.5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"elevenlabs/elevenlabs-v3": {
"id": "elevenlabs/elevenlabs-v3",
"name": "ElevenLabs-v3",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "elevenlabs",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-06-05",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/elevenlabs/elevenlabs-v3\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"elevenlabs/elevenlabs-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"elevenlabs/elevenlabs-music": {
"id": "elevenlabs/elevenlabs-music",
"name": "ElevenLabs-Music",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "elevenlabs",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-08-29",
"last_updated": "2025-08-29",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 2000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/elevenlabs/elevenlabs-music\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"elevenlabs/elevenlabs-music\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stabilityai/stablediffusionxl": {
"id": "stabilityai/stablediffusionxl",
"name": "StableDiffusionXL",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "stable-diffusion",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2023-07-09",
"last_updated": "2023-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 200,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/stabilityai/stablediffusionxl\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"stabilityai/stablediffusionxl\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"trytako/tako": {
"id": "trytako/tako",
"name": "Tako",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"family": "tako",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-08-15",
"last_updated": "2024-08-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2048,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/trytako/tako\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"trytako/tako\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ideogramai/ideogram-v2a": {
"id": "ideogramai/ideogram-v2a",
"name": "Ideogram-v2a",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "ideogram",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-02-27",
"last_updated": "2025-02-27",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 150,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/ideogramai/ideogram-v2a\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"ideogramai/ideogram-v2a\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ideogramai/ideogram-v2a-turbo": {
"id": "ideogramai/ideogram-v2a-turbo",
"name": "Ideogram-v2a-Turbo",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "ideogram",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-02-27",
"last_updated": "2025-02-27",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 150,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/ideogramai/ideogram-v2a-turbo\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"ideogramai/ideogram-v2a-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ideogramai/ideogram-v2": {
"id": "ideogramai/ideogram-v2",
"name": "Ideogram-v2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "ideogram",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-08-21",
"last_updated": "2024-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 150,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/ideogramai/ideogram-v2\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"ideogramai/ideogram-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ideogramai/ideogram": {
"id": "ideogramai/ideogram",
"name": "Ideogram",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "ideogram",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-04-03",
"last_updated": "2024-04-03",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 150,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/ideogramai/ideogram\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"ideogramai/ideogram\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.8": {
"id": "anthropic/claude-opus-4.8",
"name": "Claude-Opus-4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 128000
},
"cost": {
"input": 4.2929,
"output": 21.4646
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-opus-4.8\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-3.5": {
"id": "anthropic/claude-sonnet-3.5",
"name": "Claude-Sonnet-3.5",
"description": "Legacy model retained for compatibility with older integrations",
"family": "claude-sonnet",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-06-05",
"last_updated": "2024-06-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 189096,
"output": 8192
},
"status": "deprecated",
"cost": {
"input": 2.6,
"output": 13,
"cache_read": 0.26,
"cache_write": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-sonnet-3.5\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-3.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.7": {
"id": "anthropic/claude-opus-4.7",
"name": "Claude-Opus-4.7",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-04-15",
"last_updated": "2026-04-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 128000
},
"cost": {
"input": 4.3,
"output": 21,
"cache_read": 0.43,
"cache_write": 5.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-opus-4.7\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-3.7": {
"id": "anthropic/claude-sonnet-3.7",
"name": "Claude-Sonnet-3.7",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-02-19",
"last_updated": "2025-02-19",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 196608,
"output": 128000
},
"cost": {
"input": 2.6,
"output": 13,
"cache_read": 0.26,
"cache_write": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-sonnet-3.7\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-3.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.1": {
"id": "anthropic/claude-opus-4.1",
"name": "Claude-Opus-4.1",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 0,
"max": 31999
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 196608,
"output": 32000
},
"cost": {
"input": 13,
"output": 64,
"cache_read": 1.3,
"cache_write": 16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-opus-4.1\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-3": {
"id": "anthropic/claude-haiku-3",
"name": "Claude-Haiku-3",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-03-09",
"last_updated": "2024-03-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 189096,
"output": 8192
},
"cost": {
"input": 0.21,
"output": 1.1,
"cache_read": 0.021,
"cache_write": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-haiku-3\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.6": {
"id": "anthropic/claude-sonnet-4.6",
"name": "Claude-Sonnet-4.6",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 983040,
"output": 128000
},
"cost": {
"input": 2.6,
"output": 13,
"cache_read": 0.26,
"cache_write": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-sonnet-4.6\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-3.5": {
"id": "anthropic/claude-haiku-3.5",
"name": "Claude-Haiku-3.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-10-01",
"last_updated": "2024-10-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 189096,
"output": 8192
},
"cost": {
"input": 0.68,
"output": 3.4,
"cache_read": 0.068,
"cache_write": 0.85
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-haiku-3.5\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-3.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-4.5": {
"id": "anthropic/claude-haiku-4.5",
"name": "Claude-Haiku-4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 0,
"max": 63999
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 192000,
"output": 64000
},
"cost": {
"input": 0.85,
"output": 4.3,
"cache_read": 0.085,
"cache_write": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-haiku-4.5\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.6": {
"id": "anthropic/claude-opus-4.6",
"name": "Claude-Opus-4.6",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-02-04",
"last_updated": "2026-02-04",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 983040,
"output": 128000
},
"cost": {
"input": 4.3,
"output": 21,
"cache_read": 0.43,
"cache_write": 5.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-opus-4.6\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4": {
"id": "anthropic/claude-opus-4",
"name": "Claude-Opus-4",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-05-21",
"last_updated": "2025-05-21",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 192512,
"output": 28672
},
"cost": {
"input": 13,
"output": 64,
"cache_read": 1.3,
"cache_write": 16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-opus-4\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.5": {
"id": "anthropic/claude-sonnet-4.5",
"name": "Claude-Sonnet-4.5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 0,
"max": 31999
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-09-26",
"last_updated": "2025-09-26",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 983040,
"output": 32768
},
"cost": {
"input": 2.6,
"output": 13,
"cache_read": 0.26,
"cache_write": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-sonnet-4.5\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-3.5-june": {
"id": "anthropic/claude-sonnet-3.5-june",
"name": "Claude-Sonnet-3.5-June",
"description": "Legacy model retained for compatibility with older integrations",
"family": "claude-sonnet",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-11-18",
"last_updated": "2024-11-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 189096,
"output": 8192
},
"status": "deprecated",
"cost": {
"input": 2.6,
"output": 13,
"cache_read": 0.26,
"cache_write": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-sonnet-3.5-june\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-3.5-june\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.5": {
"id": "anthropic/claude-opus-4.5",
"name": "Claude-Opus-4.5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 0,
"max": 63999
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-11-21",
"last_updated": "2025-11-21",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 196608,
"output": 64000
},
"cost": {
"input": 4.3,
"output": 21,
"cache_read": 0.43,
"cache_write": 5.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-opus-4.5\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4": {
"id": "anthropic/claude-sonnet-4",
"name": "Claude-Sonnet-4",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-05-21",
"last_updated": "2025-05-21",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 983040,
"output": 64000
},
"cost": {
"input": 2.6,
"output": 13,
"cache_read": 0.26,
"cache_write": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/anthropic/claude-sonnet-4\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/nano-banana-pro": {
"id": "google/nano-banana-pro",
"name": "Nano-Banana-Pro",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "nano-banana",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 0
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/nano-banana-pro\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/nano-banana-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-pro": {
"id": "google/gemini-3.1-pro",
"name": "Gemini-3.1-Pro",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemini-3.1-pro\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-deep-research": {
"id": "google/gemini-deep-research",
"name": "gemini-deep-research",
"description": "Legacy model retained for compatibility with older integrations",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 0
},
"status": "deprecated",
"cost": {
"input": 1.6,
"output": 9.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemini-deep-research\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-deep-research\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.0-flash": {
"id": "google/gemini-2.0-flash",
"name": "Gemini-2.0-Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-12-11",
"last_updated": "2024-12-11",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 990000,
"output": 8192
},
"cost": {
"input": 0.1,
"output": 0.42
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemini-2.0-flash\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.0-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo-3.1-fast": {
"id": "google/veo-3.1-fast",
"name": "Veo-3.1-Fast",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/veo-3.1-fast\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/veo-3.1-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/nano-banana": {
"id": "google/nano-banana",
"name": "Nano-Banana",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "nano-banana",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 0
},
"cost": {
"input": 0.21,
"output": 1.8,
"cache_read": 0.021
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/nano-banana\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/nano-banana\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/imagen-4": {
"id": "google/imagen-4",
"name": "Imagen-4",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "imagen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/imagen-4\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/imagen-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash-lite": {
"id": "google/gemini-2.5-flash-lite",
"name": "Gemini-2.5-Flash-Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-06-19",
"last_updated": "2025-06-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1024000,
"output": 64000
},
"cost": {
"input": 0.07,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/imagen-3-fast": {
"id": "google/imagen-3-fast",
"name": "Imagen-3-Fast",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "imagen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-10-17",
"last_updated": "2024-10-17",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/imagen-3-fast\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/imagen-3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.0-flash-lite": {
"id": "google/gemini-2.0-flash-lite",
"name": "Gemini-2.0-Flash-Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-02-05",
"last_updated": "2025-02-05",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 990000,
"output": 8192
},
"cost": {
"input": 0.052,
"output": 0.21
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemini-2.0-flash-lite\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.0-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-flash-lite": {
"id": "google/gemini-3.1-flash-lite",
"name": "Gemini-3.1-Flash-Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-02-18",
"last_updated": "2026-02-18",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo-3.1": {
"id": "google/veo-3.1",
"name": "Veo-3.1",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/veo-3.1\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/veo-3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo-3-fast": {
"id": "google/veo-3-fast",
"name": "Veo-3-Fast",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-10-13",
"last_updated": "2025-10-13",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/veo-3-fast\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/veo-3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/imagen-4-fast": {
"id": "google/imagen-4-fast",
"name": "Imagen-4-Fast",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "imagen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-06-25",
"last_updated": "2025-06-25",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/imagen-4-fast\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/imagen-4-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-flash": {
"id": "google/gemini-3.5-flash",
"name": "Gemini-3.5-Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5152,
"output": 9.0909,
"cache_read": 0.1515
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemini-3.5-flash\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo-3": {
"id": "google/veo-3",
"name": "Veo-3",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-05-21",
"last_updated": "2025-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/veo-3\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/veo-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3-pro": {
"id": "google/gemini-3-pro",
"name": "Gemini-3-Pro",
"description": "Legacy model retained for compatibility with older integrations",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-10-22",
"last_updated": "2025-10-22",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"status": "deprecated",
"cost": {
"input": 1.6,
"output": 9.6,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemini-3-pro\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/lyria": {
"id": "google/lyria",
"name": "Lyria",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "lyria",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-06-04",
"last_updated": "2025-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/lyria\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/lyria\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31b": {
"id": "google/gemma-4-31b",
"name": "Gemma-4-31B",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemma-4-31b\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/imagen-4-ultra": {
"id": "google/imagen-4-ultra",
"name": "Imagen-4-Ultra",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "imagen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-05-24",
"last_updated": "2025-05-24",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/imagen-4-ultra\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/imagen-4-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo-2": {
"id": "google/veo-2",
"name": "Veo-2",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-12-02",
"last_updated": "2024-12-02",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/veo-2\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/veo-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-pro": {
"id": "google/gemini-2.5-pro",
"name": "Gemini-2.5-Pro",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 0,
"max": 32768
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-02-05",
"last_updated": "2025-02-05",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1065535,
"output": 65535
},
"cost": {
"input": 0.87,
"output": 7,
"cache_read": 0.087
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemini-2.5-pro\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3-flash": {
"id": "google/gemini-3-flash",
"name": "Gemini-3-Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-10-07",
"last_updated": "2025-10-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 2.4,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemini-3-flash\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash": {
"id": "google/gemini-2.5-flash",
"name": "Gemini-2.5-Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-04-26",
"last_updated": "2025-04-26",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1065535,
"output": 65535
},
"cost": {
"input": 0.21,
"output": 1.8,
"cache_read": 0.021
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/gemini-2.5-flash\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/imagen-3": {
"id": "google/imagen-3",
"name": "Imagen-3",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "imagen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-10-15",
"last_updated": "2024-10-15",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/google/imagen-3\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"google/imagen-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-4.7": {
"id": "novita/glm-4.7",
"name": "glm-4.7",
"description": "Legacy model retained for compatibility with older integrations",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 205000,
"output": 131072
},
"status": "deprecated",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/novita/glm-4.7\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-4.6": {
"id": "novita/glm-4.6",
"name": "GLM-4.6",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/novita/glm-4.6\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/minimax-m2.1": {
"id": "novita/minimax-m2.1",
"name": "minimax-m2.1",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-12-26",
"last_updated": "2025-12-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 205000,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/novita/minimax-m2.1\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"novita/minimax-m2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-4.6v": {
"id": "novita/glm-4.6v",
"name": "glm-4.6v",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-12-09",
"last_updated": "2025-12-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131000,
"output": 32768
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/novita/glm-4.6v\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-4.6v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/kimi-k2.6": {
"id": "novita/kimi-k2.6",
"name": "Kimi-K2.6",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-20",
"last_updated": "2026-05-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 262144
},
"cost": {
"input": 0.96,
"output": 4.04,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/novita/kimi-k2.6\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"novita/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/kimi-k2-thinking": {
"id": "novita/kimi-k2-thinking",
"name": "kimi-k2-thinking",
"description": "Kimi reasoning model for long-horizon research, planning, and tool use",
"family": "kimi-thinking",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-11-07",
"last_updated": "2025-11-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/novita/kimi-k2-thinking\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"novita/kimi-k2-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/deepseek-v3.2": {
"id": "novita/deepseek-v3.2",
"name": "DeepSeek-V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 0
},
"cost": {
"input": 0.27,
"output": 0.4,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/novita/deepseek-v3.2\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"novita/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-4.7-n": {
"id": "novita/glm-4.7-n",
"name": "glm-4.7-n",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 205000,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/novita/glm-4.7-n\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-4.7-n\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-5": {
"id": "novita/glm-5",
"name": "GLM-5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 205000,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/novita/glm-5\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/kimi-k2.5": {
"id": "novita/kimi-k2.5",
"name": "Kimi-K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/novita/kimi-k2.5\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"novita/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"novita/glm-4.7-flash": {
"id": "novita/glm-4.7-flash",
"name": "glm-4.7-flash",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 65500
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/novita/glm-4.7-flash\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"novita/glm-4.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fireworks-ai/kimi-k2.5-fw": {
"id": "fireworks-ai/kimi-k2.5-fw",
"name": "Kimi-K2.5-FW",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"input": 245760,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/fireworks-ai/kimi-k2.5-fw\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"fireworks-ai/kimi-k2.5-fw\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"lumalabs/ray2": {
"id": "lumalabs/ray2",
"name": "Ray2",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "ray",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-02-20",
"last_updated": "2025-02-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 5000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/lumalabs/ray2\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"lumalabs/ray2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"empiriolabs/deepseek-v4-flash-el": {
"id": "empiriolabs/deepseek-v4-flash-el",
"name": "DeepSeek-V4-Flash-EL",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-04-24",
"last_updated": "2026-05-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 384000
},
"cost": {
"input": 0.14,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/empiriolabs/deepseek-v4-flash-el\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"empiriolabs/deepseek-v4-flash-el\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"empiriolabs/deepseek-v4-pro-el": {
"id": "empiriolabs/deepseek-v4-pro-el",
"name": "DeepSeek-V4-Pro-EL",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-04-24",
"last_updated": "2026-05-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 384000
},
"cost": {
"input": 1.67,
"output": 3.33
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/empiriolabs/deepseek-v4-pro-el\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"empiriolabs/deepseek-v4-pro-el\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"topazlabs-co/topazlabs": {
"id": "topazlabs-co/topazlabs",
"name": "TopazLabs",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "topazlabs",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-12-03",
"last_updated": "2024-12-03",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 204,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/topazlabs-co/topazlabs\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"topazlabs-co/topazlabs\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-nano": {
"id": "openai/gpt-5-nano",
"name": "GPT-5-nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.045,
"output": 0.36,
"cache_read": 0.0045
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5-nano\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-nano": {
"id": "openai/gpt-4.1-nano",
"name": "GPT-4.1-nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-04-15",
"last_updated": "2025-04-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.09,
"output": 0.36,
"cache_read": 0.022
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-4.1-nano\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-codex": {
"id": "openai/gpt-5-codex",
"name": "GPT-5-Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.1,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5-codex\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-pro": {
"id": "openai/gpt-5-pro",
"name": "GPT-5-Pro",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 14,
"output": 110
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5-pro\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-mini-high": {
"id": "openai/o3-mini-high",
"name": "o3-mini-high",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-01-31",
"last_updated": "2025-01-31",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 0.99,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/o3-mini-high\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-mini-high\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex-mini": {
"id": "openai/gpt-5.1-codex-mini",
"name": "GPT-5.1-Codex-Mini",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-11-12",
"last_updated": "2025-11-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.22,
"output": 1.8,
"cache_read": 0.022
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.1-codex-mini\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex": {
"id": "openai/gpt-5.1-codex",
"name": "GPT-5.1-Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-11-12",
"last_updated": "2025-11-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.1,
"output": 9,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.1-codex\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/chatgpt-4o-latest": {
"id": "openai/chatgpt-4o-latest",
"name": "ChatGPT-4o-Latest",
"description": "Legacy model retained for compatibility with older integrations",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-08-14",
"last_updated": "2024-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"status": "deprecated",
"cost": {
"input": 4.5,
"output": 14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/chatgpt-4o-latest\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/chatgpt-4o-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-codex": {
"id": "openai/gpt-5.2-codex",
"name": "GPT-5.2-Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-01-14",
"last_updated": "2026-01-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.6,
"output": 13,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.2-codex\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4-classic": {
"id": "openai/gpt-4-classic",
"name": "GPT-4-Classic",
"description": "Legacy model retained for compatibility with older integrations",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-03-25",
"last_updated": "2024-03-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 4096
},
"status": "deprecated",
"cost": {
"input": 27,
"output": 54
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-4-classic\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4-classic\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-deep-research": {
"id": "openai/o3-deep-research",
"name": "o3-deep-research",
"description": "Research model for long-horizon investigation, synthesis, and analytical reports",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-06-27",
"last_updated": "2025-06-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 9,
"output": 36,
"cache_read": 2.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/o3-deep-research\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-deep-research\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/sora-2": {
"id": "openai/sora-2",
"name": "Sora-2",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "sora",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/sora-2\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/sora-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-instant": {
"id": "openai/gpt-5.3-instant",
"name": "GPT-5.3-Instant",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 111616,
"output": 16384
},
"cost": {
"input": 1.6,
"output": 13,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.3-instant\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-instant\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-pro": {
"id": "openai/gpt-5.2-pro",
"name": "GPT-5.2-Pro",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 19,
"output": 150
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.2-pro\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-codex-spark": {
"id": "openai/gpt-5.3-codex-spark",
"name": "GPT-5.3-Codex-Spark",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2026-03-04",
"last_updated": "2026-03-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.3-codex-spark\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-codex-spark\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-mini": {
"id": "openai/gpt-4.1-mini",
"name": "GPT-4.1-mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-04-15",
"last_updated": "2025-04-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.36,
"output": 1.4,
"cache_read": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-4.1-mini\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4": {
"id": "openai/gpt-5.4",
"name": "GPT-5.4",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-02-26",
"last_updated": "2026-02-26",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.2,
"output": 14,
"cache_read": 0.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.4\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4-turbo": {
"id": "openai/gpt-4-turbo",
"name": "GPT-4-Turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2023-09-13",
"last_updated": "2023-09-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 9,
"output": 27
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-4-turbo\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-3.5-turbo-raw": {
"id": "openai/gpt-3.5-turbo-raw",
"name": "GPT-3.5-Turbo-Raw",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2023-09-27",
"last_updated": "2023-09-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 4524,
"output": 2048
},
"cost": {
"input": 0.45,
"output": 1.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-3.5-turbo-raw\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-3.5-turbo-raw\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-instant": {
"id": "openai/gpt-5.2-instant",
"name": "GPT-5.2-Instant",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 1.6,
"output": 13,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.2-instant\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-instant\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/dall-e-3": {
"id": "openai/dall-e-3",
"name": "DALL-E-3",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "dall-e",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2023-11-06",
"last_updated": "2023-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 800,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/dall-e-3\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/dall-e-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1": {
"id": "openai/gpt-5.1",
"name": "GPT-5.1",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-11-12",
"last_updated": "2025-11-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.1,
"output": 9,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.1\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex-max": {
"id": "openai/gpt-5.1-codex-max",
"name": "GPT-5.1-Codex-Max",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.1,
"output": 9,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.1-codex-max\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o4-mini-deep-research": {
"id": "openai/o4-mini-deep-research",
"name": "o4-mini-deep-research",
"description": "Research model for long-horizon investigation, synthesis, and analytical reports",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-06-27",
"last_updated": "2025-06-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.8,
"output": 7.2,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/o4-mini-deep-research\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/o4-mini-deep-research\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o1": {
"id": "openai/o1",
"name": "o1",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2024-12-18",
"last_updated": "2024-12-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 14,
"output": 54
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/o1\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/o1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o": {
"id": "openai/gpt-4o",
"name": "GPT-4o",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-05-13",
"last_updated": "2024-05-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-4o\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-codex": {
"id": "openai/gpt-5.3-codex",
"name": "GPT-5.3-Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-02-10",
"last_updated": "2026-02-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.6,
"output": 13,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.3-codex\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini": {
"id": "openai/gpt-4o-mini",
"name": "GPT-4o-mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 124096,
"output": 4096
},
"cost": {
"input": 0.14,
"output": 0.54,
"cache_read": 0.068
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-4o-mini\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-image-1.5": {
"id": "openai/gpt-image-1.5",
"name": "gpt-image-1.5",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-12-16",
"last_updated": "2025-12-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-image-1.5\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-image-1.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o1-pro": {
"id": "openai/o1-pro",
"name": "o1-pro",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-03-19",
"last_updated": "2025-03-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 140,
"output": 540
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/o1-pro\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/o1-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1": {
"id": "openai/gpt-4.1",
"name": "GPT-4.1",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 1.8,
"output": 7.2,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-4.1\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-image-1": {
"id": "openai/gpt-image-1",
"name": "GPT-Image-1",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-03-31",
"last_updated": "2025-03-31",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-image-1\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-image-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-nano": {
"id": "openai/gpt-5.4-nano",
"name": "GPT-5.4-Nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.18,
"output": 1.1,
"cache_read": 0.018
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.4-nano\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-search": {
"id": "openai/gpt-4o-search",
"name": "GPT-4o-Search",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-03-11",
"last_updated": "2025-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 2.2,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-4o-search\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-search\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5-pro": {
"id": "openai/gpt-5.5-pro",
"name": "GPT-5.5-Pro",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-08",
"last_updated": "2026-04-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 27.2727,
"output": 163.6364
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.5-pro\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-image-1-mini": {
"id": "openai/gpt-image-1-mini",
"name": "GPT-Image-1-Mini",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-08-26",
"last_updated": "2025-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-image-1-mini\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-image-1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-aug": {
"id": "openai/gpt-4o-aug",
"name": "GPT-4o-Aug",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-11-21",
"last_updated": "2024-11-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 2.2,
"output": 9,
"cache_read": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-4o-aug\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-aug\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-mini": {
"id": "openai/gpt-5.4-mini",
"name": "GPT-5.4-Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-03-12",
"last_updated": "2026-03-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.68,
"output": 4,
"cache_read": 0.068
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.4-mini\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-instant": {
"id": "openai/gpt-5.1-instant",
"name": "GPT-5.1-Instant",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-11-12",
"last_updated": "2025-11-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 1.1,
"output": 9,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.1-instant\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-instant\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-image-2": {
"id": "openai/gpt-image-2",
"name": "GPT-Image-2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 5.0505,
"output": 32.3232,
"cache_read": 1.2626
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-image-2\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-image-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-3.5-turbo": {
"id": "openai/gpt-3.5-turbo",
"name": "GPT-3.5-Turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2023-09-13",
"last_updated": "2023-09-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"output": 2048
},
"cost": {
"input": 0.45,
"output": 1.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-3.5-turbo\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-3.5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-chat": {
"id": "openai/gpt-5-chat",
"name": "GPT-5-Chat",
"description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
"family": "gpt-codex",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 1.1,
"output": 9,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5-chat\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-chat\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-mini": {
"id": "openai/gpt-5-mini",
"name": "GPT-5-mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-06-25",
"last_updated": "2025-06-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.22,
"output": 1.8,
"cache_read": 0.022
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5-mini\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-pro": {
"id": "openai/gpt-5.4-pro",
"name": "GPT-5.4-Pro",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 27,
"output": 160
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.4-pro\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/sora-2-pro": {
"id": "openai/sora-2-pro",
"name": "Sora-2-Pro",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "sora",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/sora-2-pro\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/sora-2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini-search": {
"id": "openai/gpt-4o-mini-search",
"name": "GPT-4o-mini-Search",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-03-11",
"last_updated": "2025-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.14,
"output": 0.54
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-4o-mini-search\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini-search\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-3.5-turbo-instruct": {
"id": "openai/gpt-3.5-turbo-instruct",
"name": "GPT-3.5-Turbo-Instruct",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2023-09-20",
"last_updated": "2023-09-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 3500,
"output": 1024
},
"cost": {
"input": 1.4,
"output": 1.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-3.5-turbo-instruct\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-3.5-turbo-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2": {
"id": "openai/gpt-5.2",
"name": "GPT-5.2",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.6,
"output": 13,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.2\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5": {
"id": "openai/gpt-5",
"name": "GPT-5",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.1,
"output": 9,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o4-mini": {
"id": "openai/o4-mini",
"name": "o4-mini",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 0.99,
"output": 4,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/o4-mini\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/o4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4-classic-0314": {
"id": "openai/gpt-4-classic-0314",
"name": "GPT-4-Classic-0314",
"description": "Legacy model retained for compatibility with older integrations",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-08-26",
"last_updated": "2024-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 4096
},
"status": "deprecated",
"cost": {
"input": 27,
"output": 54
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-4-classic-0314\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4-classic-0314\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-mini": {
"id": "openai/o3-mini",
"name": "o3-mini",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-01-31",
"last_updated": "2025-01-31",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 0.99,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/o3-mini\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3": {
"id": "openai/o3",
"name": "o3",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.8,
"output": 7.2,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/o3\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-pro": {
"id": "openai/o3-pro",
"name": "o3-pro",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-06-10",
"last_updated": "2025-06-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 18,
"output": 72
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/o3-pro\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-08",
"last_updated": "2026-04-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 4.5455,
"output": 27.2727,
"cache_read": 0.4545
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/openai/gpt-5.5\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-3-mini": {
"id": "xai/grok-3-mini",
"name": "Grok 3 Mini",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-04-11",
"last_updated": "2025-04-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.3,
"output": 0.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/xai/grok-3-mini\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.20-multi-agent": {
"id": "xai/grok-4.20-multi-agent",
"name": "Grok-4.20-Multi-Agent",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2026-03-13",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 0
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/xai/grok-4.20-multi-agent\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4.20-multi-agent\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-code-fast-1": {
"id": "xai/grok-code-fast-1",
"name": "Grok Code Fast 1",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-08-22",
"last_updated": "2025-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.5,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/xai/grok-code-fast-1\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-code-fast-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4": {
"id": "xai/grok-4",
"name": "Grok-4",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-07-10",
"last_updated": "2025-07-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/xai/grok-4\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-3": {
"id": "xai/grok-3",
"name": "Grok 3",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-04-11",
"last_updated": "2025-04-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/xai/grok-3\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4-fast-reasoning": {
"id": "xai/grok-4-fast-reasoning",
"name": "Grok-4-Fast-Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-09-16",
"last_updated": "2025-09-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/xai/grok-4-fast-reasoning\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.1-fast-non-reasoning": {
"id": "xai/grok-4.1-fast-non-reasoning",
"name": "Grok-4.1-Fast-Non-Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/xai/grok-4.1-fast-non-reasoning\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4.1-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.1-fast-reasoning": {
"id": "xai/grok-4.1-fast-reasoning",
"name": "Grok-4.1-Fast-Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/xai/grok-4.1-fast-reasoning\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4.1-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4-fast-non-reasoning": {
"id": "xai/grok-4-fast-non-reasoning",
"name": "Grok-4-Fast-Non-Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-09-16",
"last_updated": "2025-09-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/xai/grok-4-fast-non-reasoning\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras/qwen3-32b-cs": {
"id": "cerebras/qwen3-32b-cs",
"name": "qwen3-32b-cs",
"description": "Legacy model retained for compatibility with older integrations",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-05-15",
"last_updated": "2025-05-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"status": "deprecated",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/cerebras/qwen3-32b-cs\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"cerebras/qwen3-32b-cs\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras/llama-3.1-8b-cs": {
"id": "cerebras/llama-3.1-8b-cs",
"name": "Llama-3.1-8B-CS",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-05-13",
"last_updated": "2025-05-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 0
},
"cost": {
"input": 0.1,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/cerebras/llama-3.1-8b-cs\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"cerebras/llama-3.1-8b-cs\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras/llama-3.3-70b-cs": {
"id": "cerebras/llama-3.3-70b-cs",
"name": "llama-3.3-70b-cs",
"description": "Legacy model retained for compatibility with older integrations",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-05-13",
"last_updated": "2025-05-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"status": "deprecated",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/cerebras/llama-3.3-70b-cs\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"cerebras/llama-3.3-70b-cs\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras/gpt-oss-120b-cs": {
"id": "cerebras/gpt-oss-120b-cs",
"name": "GPT-OSS-120B-CS",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"release_date": "2025-08-06",
"last_updated": "2025-08-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 0
},
"cost": {
"input": 0.35,
"output": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/cerebras/gpt-oss-120b-cs\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"cerebras/gpt-oss-120b-cs\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cerebras/qwen3-235b-2507-cs": {
"id": "cerebras/qwen3-235b-2507-cs",
"name": "qwen3-235b-2507-cs",
"description": "Legacy model retained for compatibility with older integrations",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"release_date": "2025-08-06",
"last_updated": "2025-08-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"status": "deprecated",
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/cerebras/qwen3-235b-2507-cs\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"cerebras/qwen3-235b-2507-cs\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"runwayml/runway-gen-4-turbo": {
"id": "runwayml/runway-gen-4-turbo",
"name": "Runway-Gen-4-Turbo",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "runway",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2025-05-09",
"last_updated": "2025-05-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 256,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/runwayml/runway-gen-4-turbo\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"runwayml/runway-gen-4-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"runwayml/runway": {
"id": "runwayml/runway",
"name": "Runway",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "runway",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": false,
"release_date": "2024-10-11",
"last_updated": "2024-10-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 256,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poe/runwayml/runway\", apiKey: processEnvironment[\"POE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.poe.com/v1\")!,\n apiKey: processEnvironment[\"POE_API_KEY\"]\n)\nlet session = provider.model(\"runwayml/runway\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"modelscope": {
"id": "modelscope",
"name": "ModelScope",
"baseURL": "https://api-inference.modelscope.cn/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"MODELSCOPE_API_KEY"
],
"doc": "https://modelscope.cn/docs/model-service/API-Inference/intro",
"modelCount": 7,
"models": {
"ZhipuAI/GLM-4.5": {
"id": "ZhipuAI/GLM-4.5",
"name": "GLM-4.5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelscope/ZhipuAI/GLM-4.5\", apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-inference.modelscope.cn/v1\")!,\n apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"ZhipuAI/GLM-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ZhipuAI/GLM-4.6": {
"id": "ZhipuAI/GLM-4.6",
"name": "GLM-4.6",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 98304
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelscope/ZhipuAI/GLM-4.6\", apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-inference.modelscope.cn/v1\")!,\n apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"ZhipuAI/GLM-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-30B-A3B-Instruct-2507": {
"id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
"name": "Qwen3 30B A3B Instruct 2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-30",
"last_updated": "2025-07-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelscope/Qwen/Qwen3-30B-A3B-Instruct-2507\", apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-inference.modelscope.cn/v1\")!,\n apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-30B-A3B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Coder-30B-A3B-Instruct": {
"id": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
"name": "Qwen3 Coder 30B A3B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-31",
"last_updated": "2025-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelscope/Qwen/Qwen3-Coder-30B-A3B-Instruct\", apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-inference.modelscope.cn/v1\")!,\n apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Coder-30B-A3B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B-Instruct-2507": {
"id": "Qwen/Qwen3-235B-A22B-Instruct-2507",
"name": "Qwen3 235B A22B Instruct 2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04-28",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelscope/Qwen/Qwen3-235B-A22B-Instruct-2507\", apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-inference.modelscope.cn/v1\")!,\n apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B-Thinking-2507": {
"id": "Qwen/Qwen3-235B-A22B-Thinking-2507",
"name": "Qwen3-235B-A22B-Thinking-2507",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-25",
"last_updated": "2025-07-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelscope/Qwen/Qwen3-235B-A22B-Thinking-2507\", apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-inference.modelscope.cn/v1\")!,\n apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B-Thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-30B-A3B-Thinking-2507": {
"id": "Qwen/Qwen3-30B-A3B-Thinking-2507",
"name": "Qwen3 30B A3B Thinking 2507",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-30",
"last_updated": "2025-07-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modelscope/Qwen/Qwen3-30B-A3B-Thinking-2507\", apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-inference.modelscope.cn/v1\")!,\n apiKey: processEnvironment[\"MODELSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-30B-A3B-Thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"poolside": {
"id": "poolside",
"name": "Poolside",
"baseURL": "https://inference.poolside.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"POOLSIDE_API_KEY"
],
"doc": "https://platform.poolside.ai",
"modelCount": 3,
"models": {
"poolside/laguna-xs-2.1": {
"id": "poolside/laguna-xs-2.1",
"name": "Laguna XS 2.1",
"description": "Agentic coding model from Poolside in the XS size class for local deployment",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-07-02",
"last_updated": "2026-07-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poolside/poolside/laguna-xs-2.1\", apiKey: processEnvironment[\"POOLSIDE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.poolside.ai/v1\")!,\n apiKey: processEnvironment[\"POOLSIDE_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-xs-2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-m.1": {
"id": "poolside/laguna-m.1",
"name": "Laguna M.1",
"description": "Poolside's open-weight model for agentic coding and long-horizon work",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-04-28",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poolside/poolside/laguna-m.1\", apiKey: processEnvironment[\"POOLSIDE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.poolside.ai/v1\")!,\n apiKey: processEnvironment[\"POOLSIDE_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-m.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-s-2.1": {
"id": "poolside/laguna-s-2.1",
"name": "Laguna S 2.1",
"description": "Agentic coding model from Poolside in the XS size class for local deployment",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 32768
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"poolside/poolside/laguna-s-2.1\", apiKey: processEnvironment[\"POOLSIDE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.poolside.ai/v1\")!,\n apiKey: processEnvironment[\"POOLSIDE_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-s-2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"claudinio": {
"id": "claudinio",
"name": "Claudinio",
"baseURL": "https://api.claudin.io/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"CLAUDINIO_API_KEY"
],
"doc": "https://claudin.io",
"modelCount": 2,
"models": {
"claudinio": {
"id": "claudinio",
"name": "Claudinio",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"knowledge": "2026-05",
"release_date": "2026-05-12",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0.5,
"output": 2,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"claudinio/claudinio\", apiKey: processEnvironment[\"CLAUDINIO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.claudin.io/v1\")!,\n apiKey: processEnvironment[\"CLAUDINIO_API_KEY\"]\n)\nlet session = provider.model(\"claudinio\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claudius": {
"id": "claudius",
"name": "Claudius",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"knowledge": "2026-05",
"release_date": "2026-05-12",
"last_updated": "2026-05-12",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 3,
"output": 8,
"cache_read": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"claudinio/claudius\", apiKey: processEnvironment[\"CLAUDINIO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.claudin.io/v1\")!,\n apiKey: processEnvironment[\"CLAUDINIO_API_KEY\"]\n)\nlet session = provider.model(\"claudius\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"novita-ai": {
"id": "novita-ai",
"name": "NovitaAI",
"baseURL": "https://api.novita.ai/openai",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"NOVITA_API_KEY"
],
"doc": "https://novita.ai/docs/guides/introduction",
"modelCount": 107,
"models": {
"paddlepaddle/paddleocr-vl": {
"id": "paddlepaddle/paddleocr-vl",
"name": "PaddleOCR-VL",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-10-22",
"last_updated": "2025-10-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"output": 16384
},
"cost": {
"input": 0.02,
"output": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/paddlepaddle/paddleocr-vl\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"paddlepaddle/paddleocr-vl\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.7-max": {
"id": "qwen/qwen3.7-max",
"name": "Qwen3.7-Max",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 3.75,
"cache_read": 0.25,
"cache_write": 1.5625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3.7-max\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-omni-30b-a3b-instruct": {
"id": "qwen/qwen3-omni-30b-a3b-instruct",
"name": "Qwen3 Omni 30B A3B Instruct",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-09-24",
"last_updated": "2025-09-24",
"modalities": {
"input": [
"text",
"video",
"audio",
"image"
],
"output": [
"text",
"audio"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 16384
},
"cost": {
"input": 0.25,
"output": 0.97,
"input_audio": 2.2,
"output_audio": 1.788
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-omni-30b-a3b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-omni-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-30b-a3b-fp8": {
"id": "qwen/qwen3-30b-a3b-fp8",
"name": "Qwen3 30B A3B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"release_date": "2025-04-29",
"last_updated": "2025-04-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 20000
},
"cost": {
"input": 0.09,
"output": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-30b-a3b-fp8\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-30b-a3b-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-next-80b-a3b-thinking": {
"id": "qwen/qwen3-next-80b-a3b-thinking",
"name": "Qwen3 Next 80B A3B Thinking",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-10",
"last_updated": "2025-09-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-next-80b-a3b-thinking\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-next-80b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b-thinking-2507": {
"id": "qwen/qwen3-235b-a22b-thinking-2507",
"name": "Qwen3 235B A22b Thinking 2507",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-25",
"last_updated": "2025-07-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-235b-a22b-thinking-2507\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b-thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-next-80b-a3b-instruct": {
"id": "qwen/qwen3-next-80b-a3b-instruct",
"name": "Qwen3 Next 80B A3B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-10",
"last_updated": "2025-09-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-27b": {
"id": "qwen/qwen3.5-27b",
"name": "Qwen3.5-27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-26",
"last_updated": "2026-02-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3.5-27b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-35b-a3b": {
"id": "qwen/qwen3.5-35b-a3b",
"name": "Qwen3.5-35B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-26",
"last_updated": "2026-02-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3.5-35b-a3b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b-fp8": {
"id": "qwen/qwen3-235b-a22b-fp8",
"name": "Qwen3 235B A22B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"release_date": "2025-04-29",
"last_updated": "2025-04-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 20000
},
"cost": {
"input": 0.2,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-235b-a22b-fp8\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-4b-fp8": {
"id": "qwen/qwen3-4b-fp8",
"name": "Qwen3 4B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"release_date": "2025-04-29",
"last_updated": "2025-04-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 20000
},
"cost": {
"input": 0.03,
"output": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-4b-fp8\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-4b-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen2.5-vl-72b-instruct": {
"id": "qwen/qwen2.5-vl-72b-instruct",
"name": "Qwen2.5 VL 72B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-03-25",
"last_updated": "2025-03-25",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.8,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen2.5-vl-72b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen2.5-vl-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-next": {
"id": "qwen/qwen3-coder-next",
"name": "Qwen3 Coder Next",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-03",
"last_updated": "2026-02-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.2,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-coder-next\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-30b-a3b-instruct": {
"id": "qwen/qwen3-coder-30b-a3b-instruct",
"name": "Qwen3 Coder 30b A3B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-09",
"last_updated": "2025-10-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 160000,
"output": 32768
},
"cost": {
"input": 0.07,
"output": 0.27
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-397b-a17b": {
"id": "qwen/qwen3.5-397b-a17b",
"name": "Qwen3.5-397B-A17B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-17",
"last_updated": "2026-02-17",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 64000
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-max": {
"id": "qwen/qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-24",
"last_updated": "2025-09-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 2.11,
"output": 8.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-max\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-8b-instruct": {
"id": "qwen/qwen3-vl-8b-instruct",
"name": "qwen/qwen3-vl-8b-instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-17",
"last_updated": "2025-10-17",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.08,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-vl-8b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-8b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-8b-fp8": {
"id": "qwen/qwen3-8b-fp8",
"name": "Qwen3 8B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"release_date": "2025-04-29",
"last_updated": "2025-04-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 20000
},
"cost": {
"input": 0.035,
"output": 0.138
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-8b-fp8\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-8b-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-30b-a3b-instruct": {
"id": "qwen/qwen3-vl-30b-a3b-instruct",
"name": "qwen/qwen3-vl-30b-a3b-instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-11",
"last_updated": "2025-10-11",
"modalities": {
"input": [
"text",
"video",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-vl-30b-a3b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-122b-a10b": {
"id": "qwen/qwen3.5-122b-a10b",
"name": "Qwen3.5-122B-A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-26",
"last_updated": "2026-02-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3.5-122b-a10b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-122b-a10b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-32b-fp8": {
"id": "qwen/qwen3-32b-fp8",
"name": "Qwen3 32B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"release_date": "2025-04-29",
"last_updated": "2025-04-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 20000
},
"cost": {
"input": 0.1,
"output": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-32b-fp8\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-32b-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-30b-a3b-thinking": {
"id": "qwen/qwen3-vl-30b-a3b-thinking",
"name": "qwen/qwen3-vl-30b-a3b-thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-11",
"last_updated": "2025-10-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-vl-30b-a3b-thinking\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-30b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen2.5-7b-instruct": {
"id": "qwen/qwen2.5-7b-instruct",
"name": "Qwen2.5 7B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 32000
},
"cost": {
"input": 0.07,
"output": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen2.5-7b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen2.5-7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-2.5-72b-instruct": {
"id": "qwen/qwen-2.5-72b-instruct",
"name": "Qwen 2.5 72B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-10-15",
"last_updated": "2024-10-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 8192
},
"cost": {
"input": 0.38,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen-2.5-72b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-2.5-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-235b-a22b-thinking": {
"id": "qwen/qwen3-vl-235b-a22b-thinking",
"name": "Qwen3 VL 235B A22B Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"release_date": "2025-09-24",
"last_updated": "2025-09-24",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.98,
"output": 3.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-vl-235b-a22b-thinking\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-235b-a22b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-235b-a22b-instruct": {
"id": "qwen/qwen3-vl-235b-a22b-instruct",
"name": "Qwen3 VL 235B A22B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-24",
"last_updated": "2025-09-24",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-vl-235b-a22b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-235b-a22b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b-instruct-2507": {
"id": "qwen/qwen3-235b-a22b-instruct-2507",
"name": "Qwen3 235B A22B Instruct 2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-22",
"last_updated": "2025-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.09,
"output": 0.58
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-235b-a22b-instruct-2507\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-480b-a35b-instruct": {
"id": "qwen/qwen3-coder-480b-a35b-instruct",
"name": "Qwen3 Coder 480B A35B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.38,
"output": 1.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-coder-480b-a35b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-480b-a35b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-omni-30b-a3b-thinking": {
"id": "qwen/qwen3-omni-30b-a3b-thinking",
"name": "Qwen3 Omni 30B A3B Thinking",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-24",
"last_updated": "2025-09-24",
"modalities": {
"input": [
"text",
"audio",
"video",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 16384
},
"cost": {
"input": 0.25,
"output": 0.97,
"input_audio": 2.2,
"output_audio": 1.788
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen3-omni-30b-a3b-thinking\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-omni-30b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-mt-plus": {
"id": "qwen/qwen-mt-plus",
"name": "Qwen MT Plus",
"description": "Translation model for multilingual conversion, localization, and cross-language workflows",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-09-03",
"last_updated": "2025-09-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"output": 8192
},
"cost": {
"input": 0.25,
"output": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/qwen/qwen-mt-plus\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-mt-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/ernie-4.5-300b-a47b-paddle": {
"id": "baidu/ernie-4.5-300b-a47b-paddle",
"name": "ERNIE 4.5 300B A47B",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2025-06-30",
"last_updated": "2025-06-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 123000,
"output": 12000
},
"cost": {
"input": 0.28,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/baidu/ernie-4.5-300b-a47b-paddle\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"baidu/ernie-4.5-300b-a47b-paddle\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/ernie-4.5-vl-28b-a3b": {
"id": "baidu/ernie-4.5-vl-28b-a3b",
"name": "ERNIE 4.5 VL 28B A3B",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-06-30",
"last_updated": "2026-06-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 30000,
"output": 8000
},
"cost": {
"input": 0.14,
"output": 0.56
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/baidu/ernie-4.5-vl-28b-a3b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"baidu/ernie-4.5-vl-28b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/ernie-4.5-vl-424b-a47b": {
"id": "baidu/ernie-4.5-vl-424b-a47b",
"name": "ERNIE 4.5 VL 424B A47B",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"release_date": "2025-06-30",
"last_updated": "2025-06-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 123000,
"output": 16000
},
"cost": {
"input": 0.42,
"output": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/baidu/ernie-4.5-vl-424b-a47b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"baidu/ernie-4.5-vl-424b-a47b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/ernie-4.5-21B-a3b-thinking": {
"id": "baidu/ernie-4.5-21B-a3b-thinking",
"name": "ERNIE-4.5-21B-A3B-Thinking",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"family": "ernie",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-09-19",
"last_updated": "2025-09-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0.07,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/baidu/ernie-4.5-21B-a3b-thinking\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"baidu/ernie-4.5-21B-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/ernie-4.5-21B-a3b": {
"id": "baidu/ernie-4.5-21B-a3b",
"name": "ERNIE 4.5 21B A3B",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"family": "ernie",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-06-30",
"last_updated": "2025-06-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 120000,
"output": 8000
},
"cost": {
"input": 0.07,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/baidu/ernie-4.5-21B-a3b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"baidu/ernie-4.5-21B-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/ernie-4.5-vl-28b-a3b-thinking": {
"id": "baidu/ernie-4.5-vl-28b-a3b-thinking",
"name": "ERNIE-4.5-VL-28B-A3B-Thinking",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-11-26",
"last_updated": "2025-11-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0.39,
"output": 0.39
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/baidu/ernie-4.5-vl-28b-a3b-thinking\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"baidu/ernie-4.5-vl-28b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kwaipilot/kat-coder-pro": {
"id": "kwaipilot/kat-coder-pro",
"name": "Kat Coder Pro",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01-05",
"last_updated": "2026-01-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/kwaipilot/kat-coder-pro\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"kwaipilot/kat-coder-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-nemo": {
"id": "mistralai/mistral-nemo",
"name": "Mistral Nemo",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-nemo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2024-07-30",
"last_updated": "2024-07-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 60288,
"output": 16000
},
"cost": {
"input": 0.04,
"output": 0.17
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/mistralai/mistral-nemo\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-nemo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.1": {
"id": "minimax/minimax-m2.1",
"name": "Minimax M2.1",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": false,
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/minimax/minimax-m2.1\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2": {
"id": "minimax/minimax-m2",
"name": "MiniMax-M2",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2025-10-27",
"last_updated": "2025-10-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/minimax/minimax-m2\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7-highspeed": {
"id": "minimax/minimax-m2.7-highspeed",
"name": "MiniMax-M2.7-highspeed",
"description": "Low-latency M2.7 variant for interactive coding plans and agent loops",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-05-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/minimax/minimax-m2.7-highspeed\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7": {
"id": "minimax/minimax-m2.7",
"name": "MiniMax M2.7",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax-m2.7",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/minimax/minimax-m2.7\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5": {
"id": "minimax/minimax-m2.5",
"name": "MiniMax M2.5",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 204800,
"output": 131100
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/minimax/minimax-m2.5\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5-highspeed": {
"id": "minimax/minimax-m2.5-highspeed",
"name": "MiniMax M2.5 Highspeed",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax-m2.5",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 204800,
"output": 131100
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/minimax/minimax-m2.5-highspeed\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-26b-a4b-it": {
"id": "google/gemma-4-26b-a4b-it",
"name": "Gemma 4 26B A4B",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.13,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/google/gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-27b-it": {
"id": "google/gemma-3-27b-it",
"name": "Gemma 3 27B",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-03-25",
"last_updated": "2025-03-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 98304,
"output": 16384
},
"cost": {
"input": 0.119,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/google/gemma-3-27b-it\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3-27b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31b-it": {
"id": "google/gemma-4-31b-it",
"name": "Gemma 4 31B",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/google/gemma-4-31b-it\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-12b-it": {
"id": "google/gemma-3-12b-it",
"name": "Gemma 3 12B",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-03-13",
"last_updated": "2025-03-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.05,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/google/gemma-3-12b-it\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3-12b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.7": {
"id": "zai-org/glm-4.7",
"name": "GLM-4.7",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/zai-org/glm-4.7\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.5-air": {
"id": "zai-org/glm-4.5-air",
"name": "GLM 4.5 Air",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm-air",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-10-13",
"last_updated": "2025-10-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0.13,
"output": 0.85,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/zai-org/glm-4.5-air\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-4.5-air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.6": {
"id": "zai-org/glm-4.6",
"name": "GLM 4.6",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.55,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/zai-org/glm-4.6\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.6v": {
"id": "zai-org/glm-4.6v",
"name": "GLM 4.6V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glmv",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"video",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 0.9,
"cache_read": 0.055
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/zai-org/glm-4.6v\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-4.6v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-5.2": {
"id": "zai-org/glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/zai-org/glm-5.2\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/autoglm-phone-9b-multilingual": {
"id": "zai-org/autoglm-phone-9b-multilingual",
"name": "AutoGLM-Phone-9B-Multilingual",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-10",
"last_updated": "2025-12-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.035,
"output": 0.138
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/zai-org/autoglm-phone-9b-multilingual\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/autoglm-phone-9b-multilingual\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.5": {
"id": "zai-org/glm-4.5",
"name": "GLM-4.5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/zai-org/glm-4.5\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.5v": {
"id": "zai-org/glm-4.5v",
"name": "GLM 4.5V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glmv",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-08-11",
"last_updated": "2025-08-11",
"modalities": {
"input": [
"text",
"video",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 16384
},
"cost": {
"input": 0.6,
"output": 1.8,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/zai-org/glm-4.5v\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-4.5v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-5": {
"id": "zai-org/glm-5",
"name": "GLM-5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-11",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202800,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/zai-org/glm-5\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-5.1": {
"id": "zai-org/glm-5.1",
"name": "GLM-5.1",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-27",
"last_updated": "2026-03-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 1.38,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/zai-org/glm-5.1\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.7-flash": {
"id": "zai-org/glm-4.7-flash",
"name": "GLM-4.7-Flash",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.07,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/zai-org/glm-4.7-flash\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-4.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gryphe/mythomax-l2-13b": {
"id": "gryphe/mythomax-l2-13b",
"name": "Mythomax L2 13B",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-04-25",
"last_updated": "2024-04-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 4096,
"output": 3200
},
"cost": {
"input": 0.09,
"output": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/gryphe/mythomax-l2-13b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"gryphe/mythomax-l2-13b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"microsoft/wizardlm-2-8x22b": {
"id": "microsoft/wizardlm-2-8x22b",
"name": "Wizardlm 2 8x22B",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-04-24",
"last_updated": "2024-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65535,
"output": 8000
},
"cost": {
"input": 0.62,
"output": 0.62
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/microsoft/wizardlm-2-8x22b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"microsoft/wizardlm-2-8x22b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimaxai/minimax-m1-80k": {
"id": "minimaxai/minimax-m1-80k",
"name": "MiniMax M1",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 40000
},
"cost": {
"input": 0.55,
"output": 2.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/minimaxai/minimax-m1-80k\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"minimaxai/minimax-m1-80k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-ocr": {
"id": "deepseek/deepseek-ocr",
"name": "DeepSeek-OCR",
"description": "OCR model for extracting structured text from documents and screenshots",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-24",
"last_updated": "2025-10-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0.03,
"output": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-ocr\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-ocr\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash": {
"id": "deepseek/deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393216
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-v4-flash\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-prover-v2-671b": {
"id": "deepseek/deepseek-prover-v2-671b",
"name": "Deepseek Prover V2 671B",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-04-30",
"last_updated": "2025-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 160000,
"output": 160000
},
"cost": {
"input": 0.7,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-prover-v2-671b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-prover-v2-671b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-r1-0528": {
"id": "deepseek/deepseek-r1-0528",
"name": "DeepSeek R1 0528",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-05-28",
"last_updated": "2025-05-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 32768
},
"cost": {
"input": 0.7,
"output": 2.5,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-r1-0528\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-r1-0528\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.2": {
"id": "deepseek/deepseek-v3.2",
"name": "Deepseek V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 65536
},
"cost": {
"input": 0.269,
"output": 0.4,
"cache_read": 0.1345
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-v3.2\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-r1-distill-qwen-32b": {
"id": "deepseek/deepseek-r1-distill-qwen-32b",
"name": "DeepSeek R1 Distill Qwen 32B",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-01-20",
"last_updated": "2025-01-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 64000,
"output": 32000
},
"cost": {
"input": 0.3,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-r1-distill-qwen-32b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-r1-distill-qwen-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-r1-distill-llama-70b": {
"id": "deepseek/deepseek-r1-distill-llama-70b",
"name": "DeepSeek R1 Distill LLama 70B",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2025-01-27",
"last_updated": "2025-01-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0.8,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-r1-distill-llama-70b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-r1-distill-llama-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-r1-turbo": {
"id": "deepseek/deepseek-r1-turbo",
"name": "DeepSeek R1 (Turbo)\t",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-03-05",
"last_updated": "2025-03-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 64000,
"output": 16000
},
"cost": {
"input": 0.7,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-r1-turbo\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-r1-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-r1-0528-qwen3-8b": {
"id": "deepseek/deepseek-r1-0528-qwen3-8b",
"name": "DeepSeek R1 0528 Qwen3 8B",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"release_date": "2025-05-29",
"last_updated": "2025-05-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32000
},
"cost": {
"input": 0.06,
"output": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-r1-0528-qwen3-8b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-r1-0528-qwen3-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.2-exp": {
"id": "deepseek/deepseek-v3.2-exp",
"name": "Deepseek V3.2 Exp",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 65536
},
"cost": {
"input": 0.27,
"output": 0.41
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-v3.2-exp\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.2-exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.1-terminus": {
"id": "deepseek/deepseek-v3.1-terminus",
"name": "Deepseek V3.1 Terminus",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-22",
"last_updated": "2025-09-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.27,
"output": 1,
"cache_read": 0.135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-v3.1-terminus\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.1-terminus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3-turbo": {
"id": "deepseek/deepseek-v3-turbo",
"name": "DeepSeek V3 (Turbo)\t",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-03-05",
"last_updated": "2025-03-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 64000,
"output": 16000
},
"cost": {
"input": 0.4,
"output": 1.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-v3-turbo\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3-0324": {
"id": "deepseek/deepseek-v3-0324",
"name": "DeepSeek V3 0324",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-03-25",
"last_updated": "2025-03-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 0.27,
"output": 1.12,
"cache_read": 0.135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-v3-0324\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3-0324\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro": {
"id": "deepseek/deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393216
},
"cost": {
"input": 1.6,
"output": 3.2,
"cache_read": 0.135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-v4-pro\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-ocr-2": {
"id": "deepseek/deepseek-ocr-2",
"name": "deepseek/deepseek-ocr-2",
"description": "OCR model for extracting structured text from documents and screenshots",
"attachment": true,
"reasoning": false,
"tool_call": false,
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0.03,
"output": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-ocr-2\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-ocr-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-r1-distill-qwen-14b": {
"id": "deepseek/deepseek-r1-distill-qwen-14b",
"name": "DeepSeek R1 Distill Qwen 14B",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-01-20",
"last_updated": "2025-01-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-r1-distill-qwen-14b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-r1-distill-qwen-14b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.1": {
"id": "deepseek/deepseek-v3.1",
"name": "DeepSeek V3.1",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.27,
"output": 1,
"cache_read": 0.135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/deepseek/deepseek-v3.1\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionai/ring-2.6-1t": {
"id": "inclusionai/ring-2.6-1t",
"name": "Ring-2.6-1T",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"family": "ring",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-08",
"last_updated": "2026-05-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/inclusionai/ring-2.6-1t\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"inclusionai/ring-2.6-1t\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionai/ling-2.6-1t": {
"id": "inclusionai/ling-2.6-1t",
"name": "Ling-2.6-1T",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-23",
"last_updated": "2026-06-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/inclusionai/ling-2.6-1t\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"inclusionai/ling-2.6-1t\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionai/ling-2.6-flash": {
"id": "inclusionai/ling-2.6-flash",
"name": "Ling-2.6-flash",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/inclusionai/ling-2.6-flash\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"inclusionai/ling-2.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomimimo/mimo-v2-flash": {
"id": "xiaomimimo/mimo-v2-flash",
"name": "XiaomiMiMo/MiMo-V2-Flash",
"description": "MiMo flash model for fast multimodal assistance and agent workflows",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-12-19",
"last_updated": "2025-12-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32000
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/xiaomimimo/mimo-v2-flash\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"xiaomimimo/mimo-v2-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomimimo/mimo-v2-pro": {
"id": "xiaomimimo/mimo-v2-pro",
"name": "MiMo-V2-Pro",
"description": "Earlier MiMo Pro model for multimodal agents, reasoning, and code tasks",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-03-18",
"last_updated": "2026-05-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.4,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/xiaomimimo/mimo-v2-pro\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"xiaomimimo/mimo-v2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomimimo/mimo-v2.5-pro": {
"id": "xiaomimimo/mimo-v2.5-pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-05-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.522,
"output": 1.044,
"cache_read": 0.0043,
"tiers": [
{
"input": 0.522,
"output": 1.044,
"cache_read": 0.0043,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 0.522,
"output": 1.044,
"cache_read": 0.0043
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/xiaomimimo/mimo-v2.5-pro\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"xiaomimimo/mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-3.1-8b-instruct": {
"id": "meta-llama/llama-3.1-8b-instruct",
"name": "Llama 3.1 8B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-07-24",
"last_updated": "2024-07-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"output": 16384
},
"cost": {
"input": 0.02,
"output": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/meta-llama/llama-3.1-8b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-3.1-8b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-4-maverick-17b-128e-instruct-fp8": {
"id": "meta-llama/llama-4-maverick-17b-128e-instruct-fp8",
"name": "Llama 4 Maverick Instruct",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-04-06",
"last_updated": "2025-04-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 8192
},
"cost": {
"input": 0.27,
"output": 0.85
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/meta-llama/llama-4-maverick-17b-128e-instruct-fp8\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-4-maverick-17b-128e-instruct-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-3.2-3b-instruct": {
"id": "meta-llama/llama-3.2-3b-instruct",
"name": "Llama 3.2 3B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-09-18",
"last_updated": "2024-09-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32000
},
"cost": {
"input": 0.03,
"output": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/meta-llama/llama-3.2-3b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-3.2-3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-3-8b-instruct": {
"id": "meta-llama/llama-3-8b-instruct",
"name": "Llama 3 8B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-04-25",
"last_updated": "2024-04-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0.04,
"output": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/meta-llama/llama-3-8b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-3-8b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-4-scout-17b-16e-instruct": {
"id": "meta-llama/llama-4-scout-17b-16e-instruct",
"name": "Llama 4 Scout Instruct",
"description": "Open multimodal Llama model for long-context analysis and efficient agents",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-04-06",
"last_updated": "2025-04-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.18,
"output": 0.59
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/meta-llama/llama-4-scout-17b-16e-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-4-scout-17b-16e-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-3.3-70b-instruct": {
"id": "meta-llama/llama-3.3-70b-instruct",
"name": "Llama 3.3 70B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-07",
"last_updated": "2024-12-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 120000
},
"cost": {
"input": 0.135,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/meta-llama/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/llama-3-70b-instruct": {
"id": "meta-llama/llama-3-70b-instruct",
"name": "Llama3 70B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2024-04-25",
"last_updated": "2024-04-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8000
},
"cost": {
"input": 0.51,
"output": 0.74
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/meta-llama/llama-3-70b-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/llama-3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nousresearch/hermes-2-pro-llama-3-8b": {
"id": "nousresearch/hermes-2-pro-llama-3-8b",
"name": "Hermes 2 Pro Llama 3 8B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2024-06-27",
"last_updated": "2024-06-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0.14,
"output": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/nousresearch/hermes-2-pro-llama-3-8b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"nousresearch/hermes-2-pro-llama-3-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "OpenAI: GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-06",
"last_updated": "2025-08-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.04,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/openai/gpt-oss-20b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "OpenAI GPT OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-06",
"last_updated": "2025-08-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/openai/gpt-oss-120b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sao10K/l3-70b-euryale-v2.1": {
"id": "sao10K/l3-70b-euryale-v2.1",
"name": "L3 70B Euryale V2.1\t",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-06-18",
"last_updated": "2024-06-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 1.48,
"output": 1.48
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/sao10K/l3-70b-euryale-v2.1\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"sao10K/l3-70b-euryale-v2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sao10K/l3-8b-lunaris": {
"id": "sao10K/l3-8b-lunaris",
"name": "Sao10k L3 8B Lunaris\t",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2024-11-28",
"last_updated": "2024-11-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0.05,
"output": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/sao10K/l3-8b-lunaris\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"sao10K/l3-8b-lunaris\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sao10K/L3-8B-stheno-v3.2": {
"id": "sao10K/L3-8B-stheno-v3.2",
"name": "L3 8B Stheno V3.2",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-11-29",
"last_updated": "2024-11-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 32000
},
"cost": {
"input": 0.05,
"output": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/sao10K/L3-8B-stheno-v3.2\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"sao10K/L3-8B-stheno-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sao10K/l31-70b-euryale-v2.2": {
"id": "sao10K/l31-70b-euryale-v2.2",
"name": "L31 70B Euryale V2.2",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-09-19",
"last_updated": "2024-09-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 1.48,
"output": 1.48
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/sao10K/l31-70b-euryale-v2.2\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"sao10K/l31-70b-euryale-v2.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-0905": {
"id": "moonshotai/kimi-k2-0905",
"name": "Kimi K2 0905",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-09-05",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/moonshotai/kimi-k2-0905\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-0905\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.6": {
"id": "moonshotai/kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.8,
"output": 3.4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/moonshotai/kimi-k2.6\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.7-code": {
"id": "moonshotai/kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/moonshotai/kimi-k2.7-code\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-thinking": {
"id": "moonshotai/kimi-k2-thinking",
"name": "Kimi K2 Thinking",
"description": "Kimi reasoning model for long-horizon research, planning, and tool use",
"family": "kimi-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-11-07",
"last_updated": "2026-06-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 2.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/moonshotai/kimi-k2-thinking\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k3": {
"id": "moonshotai/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/moonshotai/kimi-k3\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-instruct": {
"id": "moonshotai/kimi-k2-instruct",
"name": "Kimi K2 Instruct",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-07-11",
"last_updated": "2025-07-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.57,
"output": 2.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/moonshotai/kimi-k2-instruct\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.5": {
"id": "moonshotai/kimi-k2.5",
"name": "Kimi K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/moonshotai/kimi-k2.5\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baichuan/baichuan-m2-32b": {
"id": "baichuan/baichuan-m2-32b",
"name": "baichuan-m2-32b",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"family": "baichuan",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-08-13",
"last_updated": "2025-08-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.07,
"output": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"novita-ai/baichuan/baichuan-m2-32b\", apiKey: processEnvironment[\"NOVITA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.novita.ai/openai\")!,\n apiKey: processEnvironment[\"NOVITA_API_KEY\"]\n)\nlet session = provider.model(\"baichuan/baichuan-m2-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"nebius": {
"id": "nebius",
"name": "Nebius Token Factory",
"baseURL": "https://api.tokenfactory.nebius.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"NEBIUS_API_KEY"
],
"doc": "https://docs.tokenfactory.nebius.com/",
"modelCount": 17,
"models": {
"deepseek-ai/DeepSeek-V4-Flash-0731": {
"id": "deepseek-ai/DeepSeek-V4-Flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1024000,
"output": 1024000
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/deepseek-ai/DeepSeek-V4-Flash-0731\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro": {
"id": "deepseek-ai/DeepSeek-V4-Pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.75,
"output": 3.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/deepseek-ai/DeepSeek-V4-Pro\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Nemotron-3_5-Lightning": {
"id": "nvidia/Nemotron-3_5-Lightning",
"name": "Nemotron 3.5 Lightning 30B A3B",
"description": "Fast NVIDIA Nemotron MoE for reliable agentic tasks across enterprise workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-11",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.06,
"output": 0.24,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/nvidia/Nemotron-3_5-Lightning\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Nemotron-3_5-Lightning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Nemotron-3-Ultra-550b-a55b": {
"id": "nvidia/Nemotron-3-Ultra-550b-a55b",
"name": "Nemotron 3 Ultra 550B A55B",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1,
"output": 3,
"cache_read": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/nvidia/Nemotron-3-Ultra-550b-a55b\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Nemotron-3-Ultra-550b-a55b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-super-120b-a12b": {
"id": "nvidia/nemotron-3-super-120b-a12b",
"name": "Nemotron-3-Super-120B-A12B",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02",
"release_date": "2026-03-11",
"last_updated": "2026-03-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/nvidia/nemotron-3-super-120b-a12b\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-super-120b-a12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-27b-it": {
"id": "google/gemma-3-27b-it",
"name": "Gemma-3-27b-it",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-10",
"release_date": "2026-01-20",
"last_updated": "2026-02-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 110000,
"input": 100000,
"output": 8192
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.01,
"cache_write": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/google/gemma-3-27b-it\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3-27b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2": {
"id": "zai-org/GLM-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.4,
"output": 4.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/zai-org/GLM-5.2\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.3-Flash": {
"id": "zai-org/GLM-5.3-Flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1024000,
"output": 1024000
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/zai-org/GLM-5.3-Flash\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.3-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-30B-A3B-Instruct-2507": {
"id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
"name": "Qwen3-30B-A3B-Instruct-2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2026-01-28",
"last_updated": "2026-02-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 8192
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.01,
"cache_write": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/Qwen/Qwen3-30B-A3B-Instruct-2507\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-30B-A3B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B-Instruct-2507": {
"id": "Qwen/Qwen3-235B-A22B-Instruct-2507",
"name": "Qwen3 235B A22B Instruct 2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-07-25",
"last_updated": "2025-10-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 8192
},
"cost": {
"input": 0.2,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/Qwen/Qwen3-235B-A22B-Instruct-2507\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-397B-A17B": {
"id": "Qwen/Qwen3.5-397B-A17B",
"name": "Qwen3.5-397B-A17B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-07-15",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 250000,
"output": 8192
},
"cost": {
"input": 0.6,
"output": 3.6,
"cache_read": 0.06,
"cache_write": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/Qwen/Qwen3.5-397B-A17B\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-397B-A17B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Embedding-8B": {
"id": "Qwen/Qwen3-Embedding-8B",
"name": "Qwen3-Embedding-8B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": false,
"knowledge": "2025-10",
"release_date": "2026-01-10",
"last_updated": "2026-02-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"input": 40960,
"output": 0
},
"cost": {
"input": 0.01,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/Qwen/Qwen3-Embedding-8B\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Embedding-8B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"NousResearch/Hermes-4-405B": {
"id": "NousResearch/Hermes-4-405B",
"name": "Hermes-4-405B",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-11",
"release_date": "2026-01-30",
"last_updated": "2026-02-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 120000,
"output": 8192
},
"cost": {
"input": 1,
"output": 3,
"reasoning": 3,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/NousResearch/Hermes-4-405B\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"NousResearch/Hermes-4-405B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M3": {
"id": "MiniMaxAI/MiniMax-M3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/MiniMaxAI/MiniMax-M3\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "gpt-oss-120b",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2026-01-10",
"last_updated": "2026-02-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 124000,
"output": 8192
},
"cost": {
"input": 0.15,
"output": 0.6,
"reasoning": 0.6,
"cache_read": 0.015,
"cache_write": 0.18
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/openai/gpt-oss-120b\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.7-Code": {
"id": "moonshotai/Kimi-K2.7-Code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 8000
},
"cost": {
"input": 0.95,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/moonshotai/Kimi-K2.7-Code\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.7-Code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K3": {
"id": "moonshotai/Kimi-K3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 8000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nebius/moonshotai/Kimi-K3\", apiKey: processEnvironment[\"NEBIUS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenfactory.nebius.com/v1\")!,\n apiKey: processEnvironment[\"NEBIUS_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"minimax-cn-coding-plan": {
"id": "minimax-cn-coding-plan",
"name": "MiniMax Token Plan (minimaxi.com)",
"baseURL": "https://api.minimaxi.com/anthropic/v1",
"npm": "@ai-sdk/anthropic",
"swiftDriver": "anthropicMessages",
"env": [
"MINIMAX_API_KEY"
],
"doc": "https://platform.minimaxi.com/docs/token-plan/intro",
"modelCount": 7,
"models": {
"MiniMax-M2": {
"id": "MiniMax-M2",
"name": "MiniMax-M2",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-10-27",
"last_updated": "2025-10-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-cn-coding-plan/MiniMax-M2\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimaxi.com/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.1": {
"id": "MiniMax-M2.1",
"name": "MiniMax-M2.1",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-cn-coding-plan/MiniMax-M2.1\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimaxi.com/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.5": {
"id": "MiniMax-M2.5",
"name": "MiniMax-M2.5",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-cn-coding-plan/MiniMax-M2.5\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimaxi.com/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.5-highspeed": {
"id": "MiniMax-M2.5-highspeed",
"name": "MiniMax-M2.5-highspeed",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-13",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-cn-coding-plan/MiniMax-M2.5-highspeed\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimaxi.com/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.5-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M3": {
"id": "MiniMax-M3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal coding model for long-context reasoning and agent tasks",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-25",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 512000
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-cn-coding-plan/MiniMax-M3\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimaxi.com/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.7-highspeed": {
"id": "MiniMax-M2.7-highspeed",
"name": "MiniMax-M2.7-highspeed",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-cn-coding-plan/MiniMax-M2.7-highspeed\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimaxi.com/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.7": {
"id": "MiniMax-M2.7",
"name": "MiniMax-M2.7",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-cn-coding-plan/MiniMax-M2.7\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimaxi.com/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"xiaomi-token-plan-ams": {
"id": "xiaomi-token-plan-ams",
"name": "Xiaomi Token Plan (Europe)",
"baseURL": "https://token-plan-ams.xiaomimimo.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"XIAOMI_API_KEY"
],
"doc": "https://platform.xiaomimimo.com/#/docs",
"modelCount": 7,
"models": {
"mimo-v2.5-pro": {
"id": "mimo-v2.5-pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-ams/mimo-v2.5-pro\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-ams.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5": {
"id": "mimo-v2.5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-ams/mimo-v2.5\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-ams.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-pro": {
"id": "mimo-v2-pro",
"name": "MiMo-V2-Pro",
"description": "Earlier MiMo Pro model for multimodal agents, reasoning, and code tasks",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"status": "deprecated",
"cost": {
"input": 0,
"output": 0,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-ams/mimo-v2-pro\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-ams.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-tts-voicedesign": {
"id": "mimo-v2.5-tts-voicedesign",
"name": "MiMo-V2.5-TTS-VoiceDesign",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "mimo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-ams/mimo-v2.5-tts-voicedesign\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-ams.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-tts-voicedesign\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-tts": {
"id": "mimo-v2-tts",
"name": "MiMo-V2-TTS",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "mimo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-ams/mimo-v2-tts\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-ams.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-tts-voiceclone": {
"id": "mimo-v2.5-tts-voiceclone",
"name": "MiMo-V2.5-TTS-VoiceClone",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "mimo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-ams/mimo-v2.5-tts-voiceclone\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-ams.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-tts-voiceclone\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-tts": {
"id": "mimo-v2.5-tts",
"name": "MiMo-V2.5-TTS",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "mimo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi-token-plan-ams/mimo-v2.5-tts\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://token-plan-ams.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"zeldoc": {
"id": "zeldoc",
"name": "Zeldoc",
"baseURL": "https://api.zeldoc.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ZELDOC_API_KEY"
],
"doc": "https://docs.zeldoc.ai",
"modelCount": 1,
"models": {
"zdev": {
"id": "zdev",
"name": "ZDev",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zeldoc/zdev\", apiKey: processEnvironment[\"ZELDOC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.zeldoc.ai/v1\")!,\n apiKey: processEnvironment[\"ZELDOC_API_KEY\"]\n)\nlet session = provider.model(\"zdev\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"dinference": {
"id": "dinference",
"name": "DInference",
"baseURL": "https://api.dinference.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"DINFERENCE_API_KEY"
],
"doc": "https://dinference.com",
"modelCount": 6,
"models": {
"glm-4.7": {
"id": "glm-4.7",
"name": "GLM-4.7",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.45,
"output": 1.65
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"dinference/glm-4.7\", apiKey: processEnvironment[\"DINFERENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.dinference.com/v1\")!,\n apiKey: processEnvironment[\"DINFERENCE_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 3.89
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"dinference/glm-5.2\", apiKey: processEnvironment[\"DINFERENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.dinference.com/v1\")!,\n apiKey: processEnvironment[\"DINFERENCE_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.5": {
"id": "minimax-m2.5",
"name": "MiniMax-M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 0.22,
"output": 0.88
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"dinference/minimax-m2.5\", apiKey: processEnvironment[\"DINFERENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.dinference.com/v1\")!,\n apiKey: processEnvironment[\"DINFERENCE_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5": {
"id": "glm-5",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"dinference/glm-5\", apiKey: processEnvironment[\"DINFERENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.dinference.com/v1\")!,\n apiKey: processEnvironment[\"DINFERENCE_API_KEY\"]\n)\nlet session = provider.model(\"glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 3.89
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"dinference/glm-5.1\", apiKey: processEnvironment[\"DINFERENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.dinference.com/v1\")!,\n apiKey: processEnvironment[\"DINFERENCE_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-08",
"last_updated": "2025-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.0675,
"output": 0.27
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"dinference/gpt-oss-120b\", apiKey: processEnvironment[\"DINFERENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.dinference.com/v1\")!,\n apiKey: processEnvironment[\"DINFERENCE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"pioneer": {
"id": "pioneer",
"name": "Pioneer",
"baseURL": "https://api.pioneer.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"PIONEER_API_KEY"
],
"doc": "https://agent.pioneer.ai/llms.txt",
"modelCount": 112,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-sonnet-4-6\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro": {
"id": "gemini-3.1-pro",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gemini-3.1-pro\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"devstral-2": {
"id": "devstral-2",
"name": "Devstral 2",
"description": "Mistral's coding-agent model for repository work, terminal tasks, and software fixes",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2025-12-09",
"last_updated": "2025-12-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 131072
},
"cost": {
"input": 0.4,
"output": 2,
"cache_read": 0.4,
"cache_write": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/devstral-2\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"devstral-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-nano": {
"id": "gpt-5-nano",
"name": "GPT-5 Nano",
"description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005,
"cache_write": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-5-nano\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-max": {
"id": "qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"output": 64000
},
"cost": {
"input": 1.25,
"output": 3.75,
"cache_read": 0.25,
"cache_write": 1.5625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/qwen3.7-max\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-medium": {
"id": "mistral-medium",
"name": "Mistral Medium 3",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-medium",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-05-07",
"last_updated": "2025-05-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 64000
},
"cost": {
"input": 0.4,
"output": 2,
"cache_read": 0.4,
"cache_write": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/mistral-medium\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"mistral-medium\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ministral-3b": {
"id": "ministral-3b",
"name": "Ministral 3B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-03",
"release_date": "2024-10-16",
"last_updated": "2024-10-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4000
},
"cost": {
"input": 0.1,
"output": 0.1,
"cache_read": 0.1,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/ministral-3b\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"ministral-3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-nano": {
"id": "gpt-4.1-nano",
"name": "GPT-4.1 nano",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.05,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-4.1-nano\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-plus": {
"id": "qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.325,
"output": 1.95,
"cache_read": 0.065,
"cache_write": 0.40625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/qwen3.6-plus\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5-fast": {
"id": "claude-opus-5-fast",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-opus-5-fast\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-5-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-sol": {
"id": "gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-5.6-sol\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"devstral-small-2": {
"id": "devstral-small-2",
"name": "Devstral Small 2",
"description": "Compact multimodal coding model for repository exploration, file editing, and software agents",
"family": "devstral",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2025-12-09",
"last_updated": "2025-12-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.1,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/devstral-small-2\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"devstral-small-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5": {
"id": "claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-opus-5\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5": {
"id": "claude-opus-4-5",
"name": "Claude Opus 4.5 (latest)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-opus-4-5\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.5": {
"id": "grok-4.5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5,
"cache_write": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/grok-4.5\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-mini": {
"id": "gpt-4.1-mini",
"name": "GPT-4.1 mini",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.2,
"cache_write": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-4.1-mini\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.6-flash": {
"id": "gemini-3.6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 1.5,
"output": 7.5,
"cache_read": 0.15,
"cache_write": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gemini-3.6-flash\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-5.4\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite": {
"id": "gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65000
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.03,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1": {
"id": "gpt-5.1",
"name": "GPT-5.1",
"description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-5.1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-large-3": {
"id": "mistral-large-3",
"name": "Mistral Large 3",
"description": "Mistral's largest general model for enterprise agents, coding, and multilingual reasoning",
"family": "mistral-large",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-11",
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 131072
},
"cost": {
"input": 0.5,
"output": 1.5,
"cache_read": 0.5,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/mistral-large-3\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"mistral-large-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-opus-4-6\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash": {
"id": "gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"cache_write": 0.083333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gemini-3.5-flash\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o": {
"id": "gpt-4o",
"name": "GPT-4o",
"description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-4o\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-luna": {
"id": "gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 1,
"output": 6,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-5.6-luna\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7": {
"id": "claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-opus-4-7\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.3-codex": {
"id": "gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175,
"cache_write": 1.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-5.3-codex\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o-mini": {
"id": "gpt-4o-mini",
"name": "GPT-4o mini",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-4o-mini\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 11,
"output": 55,
"cache_read": 1.1,
"cache_write": 13.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-fable-5\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash-lite": {
"id": "gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65000
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"cache_write": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1": {
"id": "gpt-4.1",
"name": "GPT-4.1",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 1,
"cache_write": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-4.1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-3-7-sonnet-latest": {
"id": "claude-3-7-sonnet-latest",
"name": "Claude Sonnet 3.7",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-10-31",
"release_date": "2025-02-19",
"last_updated": "2025-02-19",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-3-7-sonnet-latest\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-3-7-sonnet-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-nano": {
"id": "gpt-5.4-nano",
"name": "GPT-5.4 nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-5.4-nano\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-flash": {
"id": "qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.1875,
"output": 1.125,
"cache_read": 0.0375,
"cache_write": 0.234375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/qwen3.6-flash\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-mini": {
"id": "gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075,
"cache_write": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-5.4-mini\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-max-preview": {
"id": "qwen3.6-max-preview",
"name": "Qwen3.6 Max Preview",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-20",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 240000,
"output": 64000
},
"cost": {
"input": 1.04,
"output": 6.24,
"cache_read": 0.208,
"cache_write": 1.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/qwen3.6-max-preview\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5": {
"id": "claude-haiku-4-5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-haiku-4-5\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5": {
"id": "claude-sonnet-4-5",
"name": "Claude Sonnet 4.5 (latest)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-sonnet-4-5\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1": {
"id": "claude-opus-4-1",
"name": "Claude Opus 4.1 (latest)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-opus-4-1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ministral-14b": {
"id": "ministral-14b",
"name": "Ministral 14B",
"description": "Compact multimodal Mistral model for local assistants, edge agents, and efficient tool use",
"family": "ministral",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 131072
},
"cost": {
"input": 0.2,
"output": 0.2,
"cache_read": 0.2,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/ministral-14b\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"ministral-14b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-plus": {
"id": "qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.32,
"output": 1.28,
"cache_read": 0.064,
"cache_write": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/qwen3.7-plus\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-opus-4-8\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-mini": {
"id": "gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-5-mini\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"magistral-medium": {
"id": "magistral-medium",
"name": "Magistral Medium (latest)",
"description": "Mistral reasoning model for transparent analysis, math, and complex decisions",
"family": "magistral-medium",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-06",
"release_date": "2025-03-17",
"last_updated": "2025-03-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 64000
},
"cost": {
"input": 2,
"output": 5,
"cache_read": 2,
"cache_write": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/magistral-medium\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"magistral-medium\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-terra": {
"id": "gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-5.6-terra\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-medium-3.5": {
"id": "mistral-medium-3.5",
"name": "Mistral Medium 3.5",
"description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
"family": "mistral-medium",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-29",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 131072
},
"cost": {
"input": 1.5,
"output": 7.5,
"cache_read": 1.5,
"cache_write": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/mistral-medium-3.5\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"mistral-medium-3.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-flash": {
"id": "gemini-3-flash",
"name": "Gemini 3 Flash Preview",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65000
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"cache_write": 0.083333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gemini-3-flash\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5": {
"id": "claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/claude-sonnet-5\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/gpt-5.5\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"HuggingFaceTB/SmolLM3-3B-Base": {
"id": "HuggingFaceTB/SmolLM3-3B-Base",
"name": "SmolLM3 3B Base",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2025-06-30",
"last_updated": "2025-06-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.15,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/HuggingFaceTB/SmolLM3-3B-Base\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"HuggingFaceTB/SmolLM3-3B-Base\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3": {
"id": "deepseek-ai/DeepSeek-V3",
"name": "DeepSeek-V3",
"description": "Open DeepSeek MoE chat model for coding, math, and general reasoning",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-12-26",
"last_updated": "2024-12-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 8192
},
"cost": {
"input": 0.27,
"output": 1.12,
"cache_read": 0.135,
"cache_write": 0.27
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/deepseek-ai/DeepSeek-V3\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Flash": {
"id": "deepseek-ai/DeepSeek-V4-Flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.0197,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/deepseek-ai/DeepSeek-V4-Flash\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.1": {
"id": "deepseek-ai/DeepSeek-V3.1",
"name": "DeepSeek-V3.1",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 131072
},
"cost": {
"input": 0.56,
"output": 1.68,
"cache_read": 0.56,
"cache_write": 0.56
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/deepseek-ai/DeepSeek-V3.1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro": {
"id": "deepseek-ai/DeepSeek-V4-Pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.003625,
"cache_write": 0.435
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/deepseek-ai/DeepSeek-V4-Pro\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-s-2.1": {
"id": "poolside/laguna-s-2.1",
"name": "Laguna S 2.1",
"description": "Agentic coding model from Poolside in the XS size class for local deployment",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.01,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/poolside/laguna-s-2.1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-s-2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"pioneer/auto": {
"id": "pioneer/auto",
"name": "Pioneer Auto",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"family": "auto",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2024-01-01",
"last_updated": "2025-06-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/pioneer/auto\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"pioneer/auto\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Mistral-Small-4-119B-2603": {
"id": "mistralai/Mistral-Small-4-119B-2603",
"name": "Mistral Small 4",
"description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents",
"family": "mistral-small",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-06",
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 32000
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.015,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/mistralai/Mistral-Small-4-119B-2603\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Mistral-Small-4-119B-2603\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Pixtral-12B-2409": {
"id": "mistralai/Pixtral-12B-2409",
"name": "Pixtral 12B",
"description": "Mistral vision-language model for image understanding and multimodal chat",
"family": "pixtral",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2024-09-01",
"last_updated": "2024-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4000
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.15,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/mistralai/Pixtral-12B-2409\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Pixtral-12B-2409\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Mistral-Nemo-Instruct-2407": {
"id": "mistralai/Mistral-Nemo-Instruct-2407",
"name": "Mistral Nemo",
"description": "Efficient Mistral-NVIDIA open model for multilingual chat and local deployment",
"family": "mistral-nemo",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-07-01",
"last_updated": "2024-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.02,
"output": 0.03,
"cache_read": 0.02,
"cache_write": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/mistralai/Mistral-Nemo-Instruct-2407\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Mistral-Nemo-Instruct-2407\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Codestral-22B-v0.1": {
"id": "mistralai/Codestral-22B-v0.1",
"name": "Codestral-22B-v0.1",
"description": "Open Mistral code model for fill-in-the-middle and 80+ programming languages",
"family": "codestral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-05-29",
"last_updated": "2024-05-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4000
},
"cost": {
"input": 0.3,
"output": 0.9,
"cache_read": 0.3,
"cache_write": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/mistralai/Codestral-22B-v0.1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Codestral-22B-v0.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Mistral-7B-Instruct-v0.3": {
"id": "mistralai/Mistral-7B-Instruct-v0.3",
"name": "Mistral 7B Instruct v0.3",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2023-04-30",
"last_updated": "2023-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.2,
"cache_read": 0.2,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/mistralai/Mistral-7B-Instruct-v0.3\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Mistral-7B-Instruct-v0.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Ministral-8B-Instruct-2410": {
"id": "mistralai/Ministral-8B-Instruct-2410",
"name": "Ministral 8B Instruct",
"description": "Efficient open Mistral edge model for on-device chat and function calling",
"family": "ministral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-10-16",
"last_updated": "2024-10-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4000
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.15,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/mistralai/Ministral-8B-Instruct-2410\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Ministral-8B-Instruct-2410\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Magistral-Small-2506": {
"id": "mistralai/Magistral-Small-2506",
"name": "Magistral Small",
"description": "Open Mistral reasoning model for transparent step-by-step problem solving",
"family": "magistral",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2025-06-10",
"last_updated": "2025-06-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 64000
},
"cost": {
"input": 0.5,
"output": 1.5,
"cache_read": 0.5,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/mistralai/Magistral-Small-2506\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Magistral-Small-2506\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16": {
"id": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16",
"name": "Nemotron 3 Nano 30B A3B",
"description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2025-12-15",
"last_updated": "2025-12-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.2,
"cache_read": 0.05,
"cache_write": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16": {
"id": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16",
"name": "Nemotron 3 Ultra 550B A55B",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 65000
},
"cost": {
"input": 0.5,
"output": 2.5,
"cache_read": 0.15,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8": {
"id": "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8",
"name": "Nemotron 3 Super 120B A12B",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0.09,
"output": 0.45,
"cache_read": 0.09,
"cache_write": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16": {
"id": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16",
"name": "Nemotron 3.5 Lightning 30B A3B",
"description": "Fast NVIDIA Nemotron MoE for reliable agentic tasks across enterprise workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-11",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 0.5,
"cache_read": 0.5,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-E2B-it": {
"id": "google/gemma-4-E2B-it",
"name": "Gemma 4 E2B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.1,
"cache_read": 0.1,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/google/gemma-4-E2B-it\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-E2B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31B-it": {
"id": "google/gemma-4-31B-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 0.5,
"cache_read": 0.5,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/google/gemma-4-31B-it\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-E4B-it": {
"id": "google/gemma-4-E4B-it",
"name": "Gemma 4 E4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.2,
"cache_read": 0.2,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/google/gemma-4-E4B-it\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-E4B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/diffusiongemma-26B-A4B-it": {
"id": "google/diffusiongemma-26B-A4B-it",
"name": "DiffusionGemma 26B-A4B IT",
"description": "Gemini model for general assistance, reasoning, and multimodal workflows",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-05-31",
"last_updated": "2026-05-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.5,
"output": 0.5,
"cache_read": 0.5,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/google/diffusiongemma-26B-A4B-it\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"google/diffusiongemma-26B-A4B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-12B-it": {
"id": "google/gemma-4-12B-it",
"name": "Gemma 4 12B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-05-31",
"last_updated": "2026-05-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.25,
"output": 0.25,
"cache_read": 0.25,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/google/gemma-4-12B-it\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-12B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-4b-pt": {
"id": "google/gemma-3-4b-pt",
"name": "Gemma 3 4B (Pretrained)",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2025-02-28",
"last_updated": "2025-02-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.15,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/google/gemma-3-4b-pt\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3-4b-pt\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.1": {
"id": "zai-org/GLM-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202000,
"output": 131072
},
"cost": {
"input": 0.98,
"output": 3.08,
"cache_read": 0.182,
"cache_write": 0.98
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/zai-org/GLM-5.1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2-Fast": {
"id": "zai-org/GLM-5.2-Fast",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2.1,
"output": 6.6,
"cache_read": 0.21,
"cache_write": 2.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/zai-org/GLM-5.2-Fast\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2-Fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2": {
"id": "zai-org/GLM-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1040000,
"output": 128000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26,
"cache_write": 1.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/zai-org/GLM-5.2\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"LiquidAI/LFM2-24B-A2B": {
"id": "LiquidAI/LFM2-24B-A2B",
"name": "LFM2 24B A2B",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"family": "liquid",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-01-31",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.03,
"output": 0.12,
"cache_read": 0.03,
"cache_write": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/LiquidAI/LFM2-24B-A2B\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"LiquidAI/LFM2-24B-A2B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/inkling-small": {
"id": "thinkingmachines/inkling-small",
"name": "Inkling Small",
"description": "Multimodal MoE reasoning model (276B total, 12B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-07-30",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.5,
"output": 1.2,
"cache_read": 0.1,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/thinkingmachines/inkling-small\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/inkling-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fastino/gliguard-LLMGuardrails-300M": {
"id": "fastino/gliguard-LLMGuardrails-300M",
"name": "GLiGuard LLM Guardrails 300M",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2026-04-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 4096
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.15,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/fastino/gliguard-LLMGuardrails-300M\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"fastino/gliguard-LLMGuardrails-300M\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fastino/gliner2-base-v1": {
"id": "fastino/gliner2-base-v1",
"name": "GLiNER2 Base",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-06-30",
"last_updated": "2025-06-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 4096
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.15,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/fastino/gliner2-base-v1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"fastino/gliner2-base-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fastino/gliner2-multi-v1": {
"id": "fastino/gliner2-multi-v1",
"name": "GLiNER2 Multi",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-11-30",
"last_updated": "2025-11-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 4096
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.15,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/fastino/gliner2-multi-v1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"fastino/gliner2-multi-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fastino/gliner2-multi-large-v1": {
"id": "fastino/gliner2-multi-large-v1",
"name": "GLiNER2 Multi Large",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-11-30",
"last_updated": "2025-11-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 4096
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.15,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/fastino/gliner2-multi-large-v1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"fastino/gliner2-multi-large-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fastino/gliner2-privacy-filter-PII-multi": {
"id": "fastino/gliner2-privacy-filter-PII-multi",
"name": "GLiNER2 Privacy Filter PII (Multi)",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2026-04-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 4096
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.15,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/fastino/gliner2-privacy-filter-PII-multi\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"fastino/gliner2-privacy-filter-PII-multi\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fastino/gliner2-large-v1": {
"id": "fastino/gliner2-large-v1",
"name": "GLiNER2 Large",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-06-30",
"last_updated": "2025-06-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 4096
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.15,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/fastino/gliner2-large-v1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"fastino/gliner2-large-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.1": {
"id": "meta/muse-spark-1.1",
"name": "Muse Spark 1.1",
"description": "Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-08",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/meta/muse-spark-1.1\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-8B": {
"id": "Qwen/Qwen3-8B",
"name": "Qwen3 8B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-03-31",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 40960
},
"cost": {
"input": 0.2,
"output": 0.2,
"cache_read": 0.2,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/Qwen/Qwen3-8B\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-8B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-4B-Instruct-2507": {
"id": "Qwen/Qwen3-4B-Instruct-2507",
"name": "Qwen3 4B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2025-07-31",
"last_updated": "2025-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.2,
"cache_read": 0.2,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/Qwen/Qwen3-4B-Instruct-2507\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-4B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-9B": {
"id": "Qwen/Qwen3.5-9B",
"name": "Qwen3.5 9B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 0.3,
"cache_read": 0.3,
"cache_write": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/Qwen/Qwen3.5-9B\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-9B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-1.7B-Base": {
"id": "Qwen/Qwen3-1.7B-Base",
"name": "Qwen3 1.7B Base",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2025-03-31",
"last_updated": "2025-03-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.1,
"cache_read": 0.1,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/Qwen/Qwen3-1.7B-Base\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-1.7B-Base\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B-Instruct-2507": {
"id": "Qwen/Qwen3-235B-A22B-Instruct-2507",
"name": "Qwen3 235B-A22B Instruct 2507",
"description": "Updated large open Qwen3 MoE instruct model for multilingual chat, coding, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-07-21",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 1.2,
"cache_read": 1.2,
"cache_write": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/Qwen/Qwen3-235B-A22B-Instruct-2507\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-4B-Base": {
"id": "Qwen/Qwen3-4B-Base",
"name": "Qwen3 4B Base",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2025-03-31",
"last_updated": "2025-03-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.15,
"cache_read": 0.15,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/Qwen/Qwen3-4B-Base\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-4B-Base\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-32B": {
"id": "Qwen/Qwen3-32B",
"name": "Qwen3 32B",
"description": "Dense open Qwen model for self-hosted chat, reasoning, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.9,
"output": 0.9,
"cache_read": 0.9,
"cache_write": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/Qwen/Qwen3-32B\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-32B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-27B": {
"id": "Qwen/Qwen3.6-27B",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.6,
"output": 0.6,
"cache_read": 0.6,
"cache_write": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/Qwen/Qwen3.6-27B\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen2.5-Coder-0.5B": {
"id": "Qwen/Qwen2.5-Coder-0.5B",
"name": "Qwen2.5-Coder-0.5B",
"description": "Tiny open Qwen code model for lightweight completion and on-device coding",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-11-12",
"last_updated": "2024-11-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.1,
"cache_read": 0.1,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/Qwen/Qwen2.5-Coder-0.5B\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen2.5-Coder-0.5B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-35B-A3B": {
"id": "Qwen/Qwen3.6-35B-A3B",
"name": "Qwen3.6 35B-A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 1,
"cache_read": 0.028,
"cache_write": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/Qwen/Qwen3.6-35B-A3B\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sakana/fugu-ultra": {
"id": "sakana/fugu-ultra",
"name": "Fugu Ultra",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/sakana/fugu-ultra\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"sakana/fugu-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M3": {
"id": "MiniMaxAI/MiniMax-M3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06,
"cache_write": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/MiniMaxAI/MiniMax-M3\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M2.7": {
"id": "MiniMaxAI/MiniMax-M2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131000
},
"cost": {
"input": 0.279,
"output": 1.2,
"cache_read": 0.279,
"cache_write": 0.279
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/MiniMaxAI/MiniMax-M2.7\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.2-3B": {
"id": "meta-llama/Llama-3.2-3B",
"name": "Llama-3.2-3B",
"description": "Small open Llama base model for lightweight text generation and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-09-25",
"last_updated": "2024-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.1,
"cache_read": 0.1,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/meta-llama/Llama-3.2-3B\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.2-3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.2-1B": {
"id": "meta-llama/Llama-3.2-1B",
"name": "Llama-3.2-1B",
"description": "Compact open Llama base model for lightweight and on-device use",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-09-25",
"last_updated": "2024-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.1,
"cache_read": 0.1,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/meta-llama/Llama-3.2-1B\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.2-1B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.1-8B-Instruct": {
"id": "meta-llama/Llama-3.1-8B-Instruct",
"name": "Llama 3.1 8B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2023-12-31",
"release_date": "2024-06-30",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.2,
"output": 0.2,
"cache_read": 0.2,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/meta-llama/Llama-3.1-8B-Instruct\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.1-8B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.2-3B-Instruct": {
"id": "meta-llama/Llama-3.2-3B-Instruct",
"name": "Llama 3.2 3B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2023-12-31",
"release_date": "2024-08-31",
"last_updated": "2024-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 80000
},
"cost": {
"input": 0.1,
"output": 0.335,
"cache_read": 0.1,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/meta-llama/Llama-3.2-3B-Instruct\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.2-3B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.2-1B-Instruct": {
"id": "meta-llama/Llama-3.2-1B-Instruct",
"name": "Llama 3.2 1B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2023-12-31",
"release_date": "2024-08-31",
"last_updated": "2024-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 60000
},
"cost": {
"input": 0.1,
"output": 0.201,
"cache_read": 0.1,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/meta-llama/Llama-3.2-1B-Instruct\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.2-1B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.3-70B-Instruct": {
"id": "meta-llama/Llama-3.3-70B-Instruct",
"name": "Llama-3.3-70B-Instruct",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"output": 16384
},
"cost": {
"input": 0.9,
"output": 0.9,
"cache_read": 0.9,
"cache_write": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/meta-llama/Llama-3.3-70B-Instruct\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.3-70B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.07,
"output": 0.3,
"cache_read": 0.035,
"cache_write": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/openai/gpt-oss-20b\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.015,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/openai/gpt-oss-120b\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.7-Code": {
"id": "moonshotai/Kimi-K2.7-Code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 32768
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19,
"cache_write": 0.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/moonshotai/Kimi-K2.7-Code\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.7-Code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6": {
"id": "moonshotai/Kimi-K2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 131072
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.34,
"cache_write": 0.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K3-Fast": {
"id": "moonshotai/Kimi-K3-Fast",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 4.5,
"output": 22.5,
"cache_read": 0.45,
"cache_write": 4.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/moonshotai/Kimi-K3-Fast\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K3-Fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K3": {
"id": "moonshotai/Kimi-K3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/moonshotai/Kimi-K3\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"XiaomiMiMo/MiMo-V2.5-Pro": {
"id": "XiaomiMiMo/MiMo-V2.5-Pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1050000,
"output": 131000
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.0036,
"cache_write": 0.435
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/XiaomiMiMo/MiMo-V2.5-Pro\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"XiaomiMiMo/MiMo-V2.5-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"XiaomiMiMo/MiMo-V2.5": {
"id": "XiaomiMiMo/MiMo-V2.5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1050000,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028,
"cache_write": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"pioneer/XiaomiMiMo/MiMo-V2.5\", apiKey: processEnvironment[\"PIONEER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.pioneer.ai/v1\")!,\n apiKey: processEnvironment[\"PIONEER_API_KEY\"]\n)\nlet session = provider.model(\"XiaomiMiMo/MiMo-V2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"helicone": {
"id": "helicone",
"name": "Helicone",
"baseURL": "https://ai-gateway.helicone.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"HELICONE_API_KEY"
],
"doc": "https://helicone.ai/models",
"modelCount": 90,
"models": {
"llama-3.1-8b-instruct-turbo": {
"id": "llama-3.1-8b-instruct-turbo",
"name": "Meta Llama 3.1 8B Instruct Turbo",
"description": "Compact Llama instruction model for fast chat and local deployment",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.02,
"output": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/llama-3.1-8b-instruct-turbo\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.1-8b-instruct-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-3-mini": {
"id": "grok-3-mini",
"name": "xAI Grok 3 Mini",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2024-06-01",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 0.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/grok-3-mini\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"grok-3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-nano": {
"id": "gpt-5-nano",
"name": "OpenAI GPT-5 Nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.049999999999999996,
"output": 0.39999999999999997,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-5-nano\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-1-fast-non-reasoning": {
"id": "grok-4-1-fast-non-reasoning",
"name": "xAI Grok 4.1 Fast Non-Reasoning",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-11",
"release_date": "2025-11-17",
"last_updated": "2025-11-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 0.19999999999999998,
"output": 0.5,
"cache_read": 0.049999999999999996
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/grok-4-1-fast-non-reasoning\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-1-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.1-8b-instruct": {
"id": "llama-3.1-8b-instruct",
"name": "Meta Llama 3.1 8B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"output": 16384
},
"cost": {
"input": 0.02,
"output": 0.049999999999999996
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/llama-3.1-8b-instruct\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.1-8b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-nano": {
"id": "gpt-4.1-nano",
"name": "OpenAI GPT-4.1 Nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.09999999999999999,
"output": 0.39999999999999997,
"cache_read": 0.024999999999999998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-4.1-nano\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-codex": {
"id": "gpt-5-codex",
"name": "OpenAI: GPT-5 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.12500000000000003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-5-codex\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-3-haiku-20240307": {
"id": "claude-3-haiku-20240307",
"name": "Anthropic: Claude 3 Haiku",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-03",
"release_date": "2024-03-07",
"last_updated": "2024-03-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 4096
},
"cost": {
"input": 0.25,
"output": 1.25,
"cache_read": 0.03,
"cache_write": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-3-haiku-20240307\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-3-haiku-20240307\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3": {
"id": "deepseek-v3",
"name": "DeepSeek V3",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2024-12-26",
"last_updated": "2024-12-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.56,
"output": 1.68,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/deepseek-v3\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.6": {
"id": "glm-4.6",
"name": "Zai GLM-4.6",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.44999999999999996,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/glm-4.6\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-pro": {
"id": "gpt-5-pro",
"name": "OpenAI: GPT-5 Pro",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt-pro",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 15,
"output": 120
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-5-pro\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-prompt-guard-2-86m": {
"id": "llama-prompt-guard-2-86m",
"name": "Meta Llama Prompt Guard 2 86M",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-10-01",
"last_updated": "2024-10-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 512,
"output": 2
},
"cost": {
"input": 0.01,
"output": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/llama-prompt-guard-2-86m\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"llama-prompt-guard-2-86m\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-next-80b-a3b-instruct": {
"id": "qwen3-next-80b-a3b-instruct",
"name": "Qwen3 Next 80B A3B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 16384
},
"cost": {
"input": 0.14,
"output": 1.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex-mini": {
"id": "gpt-5.1-codex-mini",
"name": "OpenAI: GPT-5.1 Codex Mini",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-codex",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.024999999999999998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-5.1-codex-mini\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-prompt-guard-2-22m": {
"id": "llama-prompt-guard-2-22m",
"name": "Meta Llama Prompt Guard 2 22M",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-10-01",
"last_updated": "2024-10-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 512,
"output": 2
},
"cost": {
"input": 0.01,
"output": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/llama-prompt-guard-2-22m\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"llama-prompt-guard-2-22m\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex": {
"id": "gpt-5.1-codex",
"name": "OpenAI: GPT-5.1 Codex",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-codex",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.12500000000000003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-5.1-codex\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-code-fast-1": {
"id": "grok-code-fast-1",
"name": "xAI Grok Code Fast 1",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2024-08-25",
"last_updated": "2024-08-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 10000
},
"cost": {
"input": 0.19999999999999998,
"output": 1.5,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/grok-code-fast-1\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"grok-code-fast-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-0905": {
"id": "kimi-k2-0905",
"name": "Kimi K2 (09/05)",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-05",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.5,
"output": 2,
"cache_read": 0.39999999999999997
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/kimi-k2-0905\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-0905\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma2-9b-it": {
"id": "gemma2-9b-it",
"name": "Google Gemma 2",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2024-06-25",
"last_updated": "2024-06-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0.01,
"output": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gemma2-9b-it\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gemma2-9b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"chatgpt-4o-latest": {
"id": "chatgpt-4o-latest",
"name": "OpenAI ChatGPT-4o",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2024-08-14",
"last_updated": "2024-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 5,
"output": 20,
"cache_read": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/chatgpt-4o-latest\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"chatgpt-4o-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite": {
"id": "gemini-2.5-flash-lite",
"name": "Google Gemini 2.5 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 512,
"max": 24576
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-07-22",
"last_updated": "2025-07-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.09999999999999999,
"output": 0.39999999999999997,
"cache_read": 0.024999999999999998,
"cache_write": 0.09999999999999999
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ernie-4.5-21b-a3b-thinking": {
"id": "ernie-4.5-21b-a3b-thinking",
"name": "Baidu Ernie 4.5 21B A3B Thinking",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"family": "ernie",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-03-16",
"last_updated": "2025-03-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8000
},
"cost": {
"input": 0.07,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/ernie-4.5-21b-a3b-thinking\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"ernie-4.5-21b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4": {
"id": "grok-4",
"name": "xAI Grok 4",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-07-09",
"last_updated": "2024-07-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/grok-4\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-thinking": {
"id": "qwen3-235b-a22b-thinking",
"name": "Qwen3 235B A22B Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-07-25",
"last_updated": "2025-07-25",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 81920
},
"cost": {
"input": 0.3,
"output": 2.9000000000000004
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/qwen3-235b-a22b-thinking\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-thinking": {
"id": "kimi-k2-thinking",
"name": "Kimi K2 Thinking",
"description": "Kimi reasoning model for long-horizon research, planning, and tool use",
"family": "kimi-thinking",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-11",
"release_date": "2025-11-06",
"last_updated": "2025-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 262144
},
"cost": {
"input": 0.48,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/kimi-k2-thinking\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-32b": {
"id": "qwen3-32b",
"name": "Qwen3 32B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04-28",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 40960
},
"cost": {
"input": 0.29,
"output": 0.59
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/qwen3-32b\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-pro-preview": {
"id": "gemini-3-pro-preview",
"name": "Google Gemini 3 Pro Preview",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-11",
"release_date": "2025-11-18",
"last_updated": "2025-11-18",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.19999999999999998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gemini-3-pro-preview\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1-20250805": {
"id": "claude-opus-4-1-20250805",
"name": "Anthropic: Claude Opus 4.1 (20250805)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024,
"max": 31999
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-opus-4-1-20250805\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-1-20250805\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-mini-2025-04-14": {
"id": "gpt-4.1-mini-2025-04-14",
"name": "OpenAI GPT-4.1 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.39999999999999997,
"output": 1.5999999999999999,
"cache_read": 0.09999999999999999
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-4.1-mini-2025-04-14\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1-mini-2025-04-14\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-mini": {
"id": "gpt-4.1-mini",
"name": "OpenAI GPT-4.1 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.39999999999999997,
"output": 1.5999999999999999,
"cache_read": 0.09999999999999999
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-4.1-mini\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar-reasoning": {
"id": "sonar-reasoning",
"name": "Perplexity Sonar Reasoning",
"description": "Web-grounded reasoning model for multi-step research and cited answers",
"family": "sonar-reasoning",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-27",
"last_updated": "2025-01-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 127000,
"output": 4096
},
"cost": {
"input": 1,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/sonar-reasoning\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"sonar-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-chat-latest": {
"id": "gpt-5-chat-latest",
"name": "OpenAI GPT-5 Chat Latest",
"description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
"family": "gpt-codex",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2024-09",
"release_date": "2024-09-30",
"last_updated": "2024-09-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.12500000000000003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-5-chat-latest\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-chat-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-reasoner": {
"id": "deepseek-reasoner",
"name": "DeepSeek Reasoner",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-20",
"last_updated": "2025-01-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 64000
},
"cost": {
"input": 0.56,
"output": 1.68,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/deepseek-reasoner\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-reasoner\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-4.5-opus": {
"id": "claude-4.5-opus",
"name": "Anthropic: Claude Opus 4.5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-11",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-4.5-opus\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-4.5-opus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-20b": {
"id": "gpt-oss-20b",
"name": "OpenAI GPT-OSS 20b",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2024-06-01",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.049999999999999996,
"output": 0.19999999999999998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-oss-20b\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-3.5-sonnet-v2": {
"id": "claude-3.5-sonnet-v2",
"name": "Anthropic: Claude 3.5 Sonnet v2",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-10-22",
"last_updated": "2024-10-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.30000000000000004,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-3.5-sonnet-v2\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-3.5-sonnet-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder": {
"id": "qwen3-coder",
"name": "Qwen3 Coder 480B A35B Instruct Turbo",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.22,
"output": 0.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/qwen3-coder\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-1-fast-reasoning": {
"id": "grok-4-1-fast-reasoning",
"name": "xAI Grok 4.1 Fast Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-11",
"release_date": "2025-11-17",
"last_updated": "2025-11-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 0.19999999999999998,
"output": 0.5,
"cache_read": 0.049999999999999996
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/grok-4-1-fast-reasoning\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-1-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-30b-a3b-instruct": {
"id": "qwen3-coder-30b-a3b-instruct",
"name": "Qwen3 Coder 30B A3B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-07-31",
"last_updated": "2025-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.09999999999999999,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1": {
"id": "gpt-5.1",
"name": "OpenAI GPT-5.1",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.12500000000000003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-5.1\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-3": {
"id": "grok-3",
"name": "xAI Grok 3",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2024-06-01",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/grok-3\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"grok-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-chat-latest": {
"id": "gpt-5.1-chat-latest",
"name": "OpenAI GPT-5.1 Chat",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-codex",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.12500000000000003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-5.1-chat-latest\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-chat-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o1-mini": {
"id": "o1-mini",
"name": "OpenAI: o1-mini",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-mini",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 65536
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/o1-mini\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"o1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-4-maverick": {
"id": "llama-4-maverick",
"name": "Meta Llama 4 Maverick 17B 128E",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/llama-4-maverick\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"llama-4-maverick\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o1": {
"id": "o1",
"name": "OpenAI: o1",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 15,
"output": 60,
"cache_read": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/o1\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"o1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o": {
"id": "gpt-4o",
"name": "OpenAI GPT-4o",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-05",
"release_date": "2024-05-13",
"last_updated": "2024-05-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-4o\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-fast-reasoning": {
"id": "grok-4-fast-reasoning",
"name": "xAI: Grok 4 Fast Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 0.19999999999999998,
"output": 0.5,
"cache_read": 0.049999999999999996
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/grok-4-fast-reasoning\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5-20250929": {
"id": "claude-sonnet-4-5-20250929",
"name": "Anthropic: Claude Sonnet 4.5 (20250929)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.30000000000000004,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-sonnet-4-5-20250929\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5-20250929\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-nemo": {
"id": "mistral-nemo",
"name": "Mistral Nemo",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-nemo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16400
},
"cost": {
"input": 20,
"output": 40
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/mistral-nemo\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"mistral-nemo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-guard-4": {
"id": "llama-guard-4",
"name": "Meta Llama Guard 4 12B",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 1024
},
"cost": {
"input": 0.21,
"output": 0.21
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/llama-guard-4\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"llama-guard-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3.2": {
"id": "deepseek-v3.2",
"name": "DeepSeek V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-22",
"last_updated": "2025-09-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 163840,
"output": 65536
},
"cost": {
"input": 0.27,
"output": 0.41
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/deepseek-v3.2\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5-20251001": {
"id": "claude-haiku-4-5-20251001",
"name": "Anthropic: Claude 4.5 Haiku (20251001)",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-10",
"release_date": "2025-10-01",
"last_updated": "2025-10-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.09999999999999999,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-haiku-4-5-20251001\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5-20251001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-tng-r1t2-chimera": {
"id": "deepseek-tng-r1t2-chimera",
"name": "DeepSeek TNG R1T2 Chimera",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-07-02",
"last_updated": "2025-07-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 130000,
"output": 163840
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/deepseek-tng-r1t2-chimera\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-tng-r1t2-chimera\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-r1-distill-llama-70b": {
"id": "deepseek-r1-distill-llama-70b",
"name": "DeepSeek R1 Distill Llama 70B",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-20",
"last_updated": "2025-01-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.03,
"output": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/deepseek-r1-distill-llama-70b\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-r1-distill-llama-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o-mini": {
"id": "gpt-4o-mini",
"name": "OpenAI GPT-4o-mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-4o-mini\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-3.5-haiku": {
"id": "claude-3.5-haiku",
"name": "Anthropic: Claude 3.5 Haiku",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-10-22",
"last_updated": "2024-10-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 0.7999999999999999,
"output": 4,
"cache_read": 0.08,
"cache_write": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-3.5-haiku\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-3.5-haiku\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"hermes-2-pro-llama-3-8b": {
"id": "hermes-2-pro-llama-3-8b",
"name": "Hermes 2 Pro Llama 3 8B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-05",
"release_date": "2024-05-27",
"last_updated": "2024-05-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/hermes-2-pro-llama-3-8b\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"hermes-2-pro-llama-3-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1": {
"id": "gpt-4.1",
"name": "OpenAI GPT-4.1",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-4.1\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar": {
"id": "sonar",
"name": "Perplexity Sonar",
"description": "Sonar search model for current answers, retrieval, and citation-backed chat",
"family": "sonar",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-27",
"last_updated": "2025-01-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 127000,
"output": 4096
},
"cost": {
"input": 1,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/sonar\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"sonar\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-0711": {
"id": "kimi-k2-0711",
"name": "Kimi K2 (07/11)",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.5700000000000001,
"output": 2.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/kimi-k2-0711\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-0711\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar-reasoning-pro": {
"id": "sonar-reasoning-pro",
"name": "Perplexity Sonar Reasoning Pro",
"description": "Web-grounded reasoning model for multi-step research and cited answers",
"family": "sonar-reasoning",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-27",
"last_updated": "2025-01-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 127000,
"output": 4096
},
"cost": {
"input": 2,
"output": 8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/sonar-reasoning-pro\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"sonar-reasoning-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4": {
"id": "claude-opus-4",
"name": "Anthropic: Claude Opus 4",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024,
"max": 31999
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-05-14",
"last_updated": "2025-05-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-opus-4\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-30b-a3b": {
"id": "qwen3-30b-a3b",
"name": "Qwen3 30B A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2025-06-01",
"last_updated": "2025-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 41000,
"output": 41000
},
"cost": {
"input": 0.08,
"output": 0.29
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/qwen3-30b-a3b\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-4-scout": {
"id": "llama-4-scout",
"name": "Meta Llama 4 Scout 17B 16E",
"description": "Open multimodal Llama model for long-context analysis and efficient agents",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.08,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/llama-4-scout\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"llama-4-scout\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3.1-terminus": {
"id": "deepseek-v3.1-terminus",
"name": "DeepSeek V3.1 Terminus",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-22",
"last_updated": "2025-09-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.27,
"output": 1,
"cache_read": 0.21600000000000003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/deepseek-v3.1-terminus\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3.1-terminus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1": {
"id": "claude-opus-4-1",
"name": "Anthropic: Claude Opus 4.1",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024,
"max": 31999
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-opus-4-1\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-3.7-sonnet": {
"id": "claude-3.7-sonnet",
"name": "Anthropic: Claude 3.7 Sonnet",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-02",
"release_date": "2025-02-19",
"last_updated": "2025-02-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.30000000000000004,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-3.7-sonnet\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-3.7-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small": {
"id": "mistral-small",
"name": "Mistral Small 3.2",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-06-20",
"last_updated": "2025-06-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.075,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/mistral-small\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-large-2411": {
"id": "mistral-large-2411",
"name": "Mistral-Large",
"description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
"family": "mistral-large",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-07-24",
"last_updated": "2024-07-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 2,
"output": 6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/mistral-large-2411\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"mistral-large-2411\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-235b-a22b-instruct": {
"id": "qwen3-vl-235b-a22b-instruct",
"name": "Qwen3 VL 235B A22B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0.3,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/qwen3-vl-235b-a22b-instruct\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-235b-a22b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar-pro": {
"id": "sonar-pro",
"name": "Perplexity Sonar Pro",
"description": "Advanced Sonar search model for deeper research and cited synthesis",
"family": "sonar-pro",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-27",
"last_updated": "2025-01-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 4096
},
"cost": {
"input": 3,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/sonar-pro\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"sonar-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-4.5-sonnet": {
"id": "claude-4.5-sonnet",
"name": "Anthropic: Claude Sonnet 4.5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.30000000000000004,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-4.5-sonnet\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-4.5-sonnet\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-mini": {
"id": "gpt-5-mini",
"name": "OpenAI GPT-5 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.024999999999999998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-5-mini\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "OpenAI GPT-OSS 120b",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2024-06-01",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.04,
"output": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-oss-120b\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar-deep-research": {
"id": "sonar-deep-research",
"name": "Perplexity Sonar Deep Research",
"description": "Sonar search model for current answers, retrieval, and citation-backed chat",
"family": "sonar-deep-research",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-01-27",
"last_updated": "2025-01-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 127000,
"output": 4096
},
"cost": {
"input": 2,
"output": 8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/sonar-deep-research\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"sonar-deep-research\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.1-8b-instant": {
"id": "llama-3.1-8b-instant",
"name": "Meta Llama 3.1 8B Instant",
"description": "Compact Llama instruction model for fast chat and local deployment",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2024-07-01",
"last_updated": "2024-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32678
},
"cost": {
"input": 0.049999999999999996,
"output": 0.08
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/llama-3.1-8b-instant\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.1-8b-instant\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "Google Gemini 2.5 Pro",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.3125,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gemini-2.5-pro\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2.5-coder-7b-fast": {
"id": "qwen2.5-coder-7b-fast",
"name": "Qwen2.5 Coder 7B fast",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2024-09-15",
"last_updated": "2024-09-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 8192
},
"cost": {
"input": 0.03,
"output": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/qwen2.5-coder-7b-fast\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2.5-coder-7b-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-4.5-haiku": {
"id": "claude-4.5-haiku",
"name": "Anthropic: Claude 4.5 Haiku",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-10",
"release_date": "2025-10-01",
"last_updated": "2025-10-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.09999999999999999,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-4.5-haiku\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-4.5-haiku\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5": {
"id": "gpt-5",
"name": "OpenAI GPT-5",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.12500000000000003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gpt-5\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "Google Gemini 2.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.075,
"cache_write": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gemini-2.5-flash\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4": {
"id": "claude-sonnet-4",
"name": "Anthropic: Claude Sonnet 4",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-05-14",
"last_updated": "2025-05-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.30000000000000004,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/claude-sonnet-4\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-3-12b-it": {
"id": "gemma-3-12b-it",
"name": "Google Gemma 3 12B",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2024-12-01",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.049999999999999996,
"output": 0.09999999999999999
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/gemma-3-12b-it\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"gemma-3-12b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-70b-versatile": {
"id": "llama-3.3-70b-versatile",
"name": "Meta Llama 3.3 70B Versatile",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32678
},
"cost": {
"input": 0.59,
"output": 0.7899999999999999
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/llama-3.3-70b-versatile\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.3-70b-versatile\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-70b-instruct": {
"id": "llama-3.3-70b-instruct",
"name": "Meta Llama 3.3 70B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16400
},
"cost": {
"input": 0.13,
"output": 0.39
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-fast-non-reasoning": {
"id": "grok-4-fast-non-reasoning",
"name": "xAI Grok 4 Fast Non-Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-19",
"last_updated": "2025-09-19",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 0.19999999999999998,
"output": 0.5,
"cache_read": 0.049999999999999996
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/grok-4-fast-non-reasoning\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o4-mini": {
"id": "o4-mini",
"name": "OpenAI o4 Mini",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-mini",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2024-06",
"release_date": "2024-06-01",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/o4-mini\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"o4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o3-mini": {
"id": "o3-mini",
"name": "OpenAI o3 Mini",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-mini",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2023-10",
"release_date": "2023-10-01",
"last_updated": "2023-10-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/o3-mini\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o3": {
"id": "o3",
"name": "OpenAI o3",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2024-06",
"release_date": "2024-06-01",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/o3\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o3-pro": {
"id": "o3-pro",
"name": "OpenAI o3 Pro",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o-pro",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": false,
"knowledge": "2024-06",
"release_date": "2024-06-01",
"last_updated": "2024-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 20,
"output": 80
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"helicone/o3-pro\", apiKey: processEnvironment[\"HELICONE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://ai-gateway.helicone.ai/v1\")!,\n apiKey: processEnvironment[\"HELICONE_API_KEY\"]\n)\nlet session = provider.model(\"o3-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"cloudferro-sherlock": {
"id": "cloudferro-sherlock",
"name": "CloudFerro Sherlock",
"baseURL": "https://api-sherlock.cloudferro.com/openai/v1/",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"CLOUDFERRO_SHERLOCK_API_KEY"
],
"doc": "https://docs.sherlock.cloudferro.com/",
"modelCount": 5,
"models": {
"MiniMaxAI/MiniMax-M2.5": {
"id": "MiniMaxAI/MiniMax-M2.5",
"name": "MiniMax-M2.5",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-01",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196000,
"input": 180000,
"output": 16000
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudferro-sherlock/MiniMaxAI/MiniMax-M2.5\", apiKey: processEnvironment[\"CLOUDFERRO_SHERLOCK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-sherlock.cloudferro.com/openai/v1/\")!,\n apiKey: processEnvironment[\"CLOUDFERRO_SHERLOCK_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.3-70B-Instruct": {
"id": "meta-llama/Llama-3.3-70B-Instruct",
"name": "Llama 3.3 70B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10-09",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 70000,
"output": 70000
},
"cost": {
"input": 2.92,
"output": 2.92
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudferro-sherlock/meta-llama/Llama-3.3-70B-Instruct\", apiKey: processEnvironment[\"CLOUDFERRO_SHERLOCK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-sherlock.cloudferro.com/openai/v1/\")!,\n apiKey: processEnvironment[\"CLOUDFERRO_SHERLOCK_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.3-70B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "OpenAI GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-28",
"last_updated": "2025-08-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131000,
"output": 131000
},
"cost": {
"input": 2.92,
"output": 2.92
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudferro-sherlock/openai/gpt-oss-120b\", apiKey: processEnvironment[\"CLOUDFERRO_SHERLOCK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-sherlock.cloudferro.com/openai/v1/\")!,\n apiKey: processEnvironment[\"CLOUDFERRO_SHERLOCK_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"speakleash/Bielik-11B-v2.6-Instruct": {
"id": "speakleash/Bielik-11B-v2.6-Instruct",
"name": "Bielik 11B v2.6 Instruct",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-03-13",
"last_updated": "2025-03-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 32000
},
"cost": {
"input": 0.67,
"output": 0.67
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudferro-sherlock/speakleash/Bielik-11B-v2.6-Instruct\", apiKey: processEnvironment[\"CLOUDFERRO_SHERLOCK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-sherlock.cloudferro.com/openai/v1/\")!,\n apiKey: processEnvironment[\"CLOUDFERRO_SHERLOCK_API_KEY\"]\n)\nlet session = provider.model(\"speakleash/Bielik-11B-v2.6-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"speakleash/Bielik-11B-v3.0-Instruct": {
"id": "speakleash/Bielik-11B-v3.0-Instruct",
"name": "Bielik 11B v3.0 Instruct",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-03-13",
"last_updated": "2025-03-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 32000
},
"cost": {
"input": 0.67,
"output": 0.67
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudferro-sherlock/speakleash/Bielik-11B-v3.0-Instruct\", apiKey: processEnvironment[\"CLOUDFERRO_SHERLOCK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-sherlock.cloudferro.com/openai/v1/\")!,\n apiKey: processEnvironment[\"CLOUDFERRO_SHERLOCK_API_KEY\"]\n)\nlet session = provider.model(\"speakleash/Bielik-11B-v3.0-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"stepfun": {
"id": "stepfun",
"name": "StepFun (China)",
"baseURL": "https://api.stepfun.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"STEPFUN_API_KEY"
],
"doc": "https://platform.stepfun.com/docs/zh/overview/concept",
"modelCount": 8,
"models": {
"step-3.7-flash": {
"id": "step-3.7-flash",
"name": "Step 3.7 Flash",
"description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-06-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"cost": {
"input": 0.185,
"output": 1.11,
"cache_read": 0.037
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun/step-3.7-flash\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.com/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-tts-2": {
"id": "step-tts-2",
"name": "Step TTS 2",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "step",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-03-01",
"last_updated": "2026-07-02",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun/step-tts-2\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.com/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-tts-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3.5-flash": {
"id": "step-3.5-flash",
"name": "Step 3.5 Flash",
"description": "StepFun flash lane for quick multimodal reasoning and coding assistance",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-29",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun/step-3.5-flash\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.com/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-2-16k": {
"id": "step-2-16k",
"name": "Step 2 (16K)",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2025-01-01",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"input": 16384,
"output": 8192
},
"cost": {
"input": 5.21,
"output": 16.44,
"cache_read": 1.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun/step-2-16k\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.com/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-2-16k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepaudio-2.5-asr": {
"id": "stepaudio-2.5-asr",
"name": "StepAudio 2.5 ASR",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "step",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-04-24",
"last_updated": "2026-07-02",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun/stepaudio-2.5-asr\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.com/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"stepaudio-2.5-asr\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3.5-flash-2603": {
"id": "step-3.5-flash-2603",
"name": "Step 3.5 Flash 2603",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun/step-3.5-flash-2603\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.com/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-3.5-flash-2603\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepaudio-2.5-tts": {
"id": "stepaudio-2.5-tts",
"name": "StepAudio 2.5 TTS",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "step",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-04-16",
"last_updated": "2026-07-02",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun/stepaudio-2.5-tts\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.com/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"stepaudio-2.5-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-1-32k": {
"id": "step-1-32k",
"name": "Step 1 (32K)",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2025-01-01",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 2.05,
"output": 9.59,
"cache_read": 0.41
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stepfun/step-1-32k\", apiKey: processEnvironment[\"STEPFUN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stepfun.com/v1\")!,\n apiKey: processEnvironment[\"STEPFUN_API_KEY\"]\n)\nlet session = provider.model(\"step-1-32k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"unorouter": {
"id": "unorouter",
"name": "UnoRouter",
"baseURL": "https://api.unorouter.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"UNOROUTER_API_KEY"
],
"doc": "https://unorouter.com/models",
"modelCount": 23,
"models": {
"deepseek-v4-pro:free": {
"id": "deepseek-v4-pro:free",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/deepseek-v4-pro:free\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.7": {
"id": "minimax-m2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.819,
"output": 3.276
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/minimax-m2.7\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash:free": {
"id": "deepseek-v4-flash:free",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/deepseek-v4-flash:free\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.2675,
"output": 5.3368
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/kimi-k2.6\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.6001,
"output": 5.0288
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/glm-5.2\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-397b-a17b:free": {
"id": "qwen3.5-397b-a17b:free",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/qwen3.5-397b-a17b:free\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-397b-a17b:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.0625,
"output": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/deepseek-v4-flash\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 1.8,
"output": 10.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/gpt-5.4\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash": {
"id": "gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.1857,
"output": 1.1142
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/gemini-3.5-flash\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.5-flash:free": {
"id": "glm-4.5-flash:free",
"name": "GLM-4.5-Flash",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/glm-4.5-flash:free\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5-flash:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5-20251001": {
"id": "claude-haiku-4-5-20251001",
"name": "Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1.2,
"output": 6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/claude-haiku-4-5-20251001\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5-20251001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3.7-flash:free": {
"id": "step-3.7-flash:free",
"name": "Step 3.7 Flash",
"description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/step-3.7-flash:free\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"step-3.7-flash:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-ultra-550b-a55b:free": {
"id": "nemotron-3-ultra-550b-a55b:free",
"name": "Nemotron 3 Ultra 550B A55B",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/nemotron-3-ultra-550b-a55b:free\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-ultra-550b-a55b:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it:free": {
"id": "gemma-4-31b-it:free",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/gemma-4-31b-it:free\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2:free": {
"id": "glm-5.2:free",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/glm-5.2:free\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.7:free": {
"id": "minimax-m2.7:free",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/minimax-m2.7:free\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.7:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4:free": {
"id": "gpt-5.4:free",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/gpt-5.4:free\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5:free": {
"id": "gpt-5.5:free",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/gpt-5.5:free\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.425,
"output": 2.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/claude-opus-4-8\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.8999,
"output": 1.7999
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/deepseek-v4-pro\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2": {
"id": "gpt-5.2",
"name": "GPT-5.2",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.05,
"output": 8.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/gpt-5.2\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5": {
"id": "claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.44,
"output": 7.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/claude-sonnet-5\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.1875,
"output": 1.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"unorouter/gpt-5.5\", apiKey: processEnvironment[\"UNOROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.unorouter.com/v1\")!,\n apiKey: processEnvironment[\"UNOROUTER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"coralbricks": {
"id": "coralbricks",
"name": "CoralBricks",
"baseURL": "https://inference.coralbricks.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"CORAL_API_KEY"
],
"doc": "https://www.coralbricks.ai/docs",
"modelCount": 3,
"models": {
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"coralbricks/kimi-k3\", apiKey: processEnvironment[\"CORAL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.coralbricks.ai/v1\")!,\n apiKey: processEnvironment[\"CORAL_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-fp4": {
"id": "glm-5.3-fp4",
"name": "GLM 5.3 FP4",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.12,
"output": 4.4,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"coralbricks/glm-5.3-fp4\", apiKey: processEnvironment[\"CORAL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.coralbricks.ai/v1\")!,\n apiKey: processEnvironment[\"CORAL_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-fp4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.12,
"output": 0.6,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"coralbricks/gpt-oss-120b\", apiKey: processEnvironment[\"CORAL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.coralbricks.ai/v1\")!,\n apiKey: processEnvironment[\"CORAL_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"hyper": {
"id": "hyper",
"name": "Charm Hyper",
"baseURL": "https://hyper.charm.land/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"HYPER_API_KEY"
],
"doc": "https://hyper.charm.land",
"modelCount": 34,
"models": {
"qwen3.7-max": {
"id": "qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-28",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3.7-max\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-it": {
"id": "gemma-4-26b-a4b-it",
"name": "Gemma 4 26B A4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-30",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 25600
},
"cost": {
"input": 0.102,
"output": 0.356,
"cache_read": 0.051
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro-0813": {
"id": "deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 1.437216,
"output": 4.311648,
"cache_read": 0.047907
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-08-02",
"last_updated": "2026-08-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.44,
"output": 1.32,
"cache_read": 0.044
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-next-80b-a3b-instruct": {
"id": "qwen3-next-80b-a3b-instruct",
"name": "Qwen3-Next 80B-A3B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-30",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 26214
},
"cost": {
"input": 0.1175,
"output": 1.136,
"cache_read": 0.05875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-plus": {
"id": "qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-05-20",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3.6-plus\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-27b": {
"id": "qwen3.8-27b",
"name": "Qwen3.8 27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3.8-27b\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.7": {
"id": "minimax-m2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-05",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262100,
"output": 6553
},
"cost": {
"input": 0.396,
"output": 1.464,
"cache_read": 0.198
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/minimax-m2.7\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-07-03",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 26214
},
"cost": {
"input": 1.03436,
"output": 4.3552,
"cache_read": 0.174208
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/kimi-k2.6\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-4-maverick-17b-128e-instruct-fp8": {
"id": "llama-4-maverick-17b-128e-instruct-fp8",
"name": "Llama 4 Maverick 17B Instruct",
"description": "Open multimodal Llama for strong reasoning with efficient everyday serving",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2026-04-30",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 430000,
"output": 43000
},
"cost": {
"input": 0.255,
"output": 0.8365,
"cache_read": 0.1275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/llama-4-maverick-17b-128e-instruct-fp8\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"llama-4-maverick-17b-128e-instruct-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-30",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 1.52432,
"output": 4.79072,
"cache_read": 0.152432
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/glm-5.2\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m3": {
"id": "minimax-m3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-30",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512000,
"output": 512000
},
"cost": {
"input": 0.32664,
"output": 1.30656,
"cache_read": 0.064239
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/minimax-m3\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-06",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.2,
"output": 0.4,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/deepseek-v4-flash\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code": {
"id": "kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-07-03",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 16000
},
"cost": {
"input": 1.03436,
"output": 4.3552,
"cache_read": 0.206872
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/kimi-k2.7-code\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-thinking": {
"id": "kimi-k2-thinking",
"name": "Kimi K2 Thinking",
"description": "Thinking Kimi model for slower research passes, planning, and hard technical questions",
"family": "kimi-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 26214
},
"cost": {
"input": 0.6,
"output": 2.5,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/kimi-k2-thinking\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4.1-flash": {
"id": "deepseek-v4.1-flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 26214
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/deepseek-v4.1-flash\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4.1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-flash": {
"id": "qwen3.7-flash",
"name": "Qwen3.7 Flash",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-27",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 991000,
"output": 64000
},
"cost": {
"input": 0.2,
"output": 0.8,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3.7-flash\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-27",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 16000
},
"cost": {
"input": 3.2664,
"output": 16.332,
"cache_read": 0.32664
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/kimi-k3\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-31",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.16332,
"output": 0.5444,
"cache_read": 0.031575
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/glm-5.3-flash\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-2.4t-a95b": {
"id": "qwen3.8-2.4t-a95b",
"name": "Qwen3.8 2.4T A95B",
"description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3.8-2.4t-a95b\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-2.4t-a95b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-flash": {
"id": "qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-20",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 1,
"output": 4,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3.6-flash\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-flash": {
"id": "qwen3.8-flash",
"name": "Qwen3.8 Flash",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.15,
"output": 0.47,
"cache_read": 0.016
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3.8-flash\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inkling": {
"id": "inkling",
"name": "Inkling",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-08-15",
"last_updated": "2026-09-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 32768
},
"cost": {
"input": 1.0888,
"output": 4.40964,
"cache_read": 0.185096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/inkling\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5": {
"id": "glm-5",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-04-13",
"last_updated": "2026-08-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 20275
},
"cost": {
"input": 0.86,
"output": 2.752,
"cache_read": 0.43
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/glm-5\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max": {
"id": "qwen3.8-max",
"name": "Qwen3.8 Max Preview",
"description": "Preview Qwen flagship for million-token multimodal reasoning and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-19",
"last_updated": "2026-07-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3.8-max\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-04-13",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 26214
},
"cost": {
"input": 0.5584,
"output": 2.935,
"cache_read": 0.2792
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/kimi-k2.5\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202750,
"output": 3276
},
"cost": {
"input": 1.318,
"output": 4.308,
"cache_read": 0.659
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/glm-5.1\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-plus": {
"id": "qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-15",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 1.2,
"output": 4.8,
"cache_read": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3.7-plus\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-06",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 2.4,
"output": 4.8,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/deepseek-v4-pro\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-13",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128072,
"output": 13107
},
"cost": {
"input": 0.168,
"output": 0.66,
"cache_read": 0.084
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/gpt-oss-120b\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-28",
"last_updated": "2026-08-31",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 1.52432,
"output": 4.79072,
"cache_read": 0.283088
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/glm-5.3\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-70b-instruct": {
"id": "llama-3.3-70b-instruct",
"name": "Llama-3.3-70B-Instruct",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2026-04-30",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 12800
},
"cost": {
"input": 0.6066,
"output": 1.0386,
"cache_read": 0.3033
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-max": {
"id": "qwen3.6-max",
"name": "Qwen3.6 Max Preview",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-05-20",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3.6-max\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-480b-a35b-instruct-int4-mixed-ar": {
"id": "qwen3-coder-480b-a35b-instruct-int4-mixed-ar",
"name": "Qwen3-Coder 480B-A35B Instruct",
"description": "Open Qwen coding heavyweight for repository reasoning and agentic engineering",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-30",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 106000,
"output": 10600
},
"cost": {
"input": 0.445,
"output": 2.145,
"cache_read": 0.2225
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"hyper/qwen3-coder-480b-a35b-instruct-int4-mixed-ar\", apiKey: processEnvironment[\"HYPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://hyper.charm.land/v1\")!,\n apiKey: processEnvironment[\"HYPER_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-480b-a35b-instruct-int4-mixed-ar\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"requesty": {
"id": "requesty",
"name": "Requesty",
"baseURL": "https://router.requesty.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"REQUESTY_API_KEY"
],
"doc": "https://requesty.ai/solution/llm-routing/models",
"modelCount": 154,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-sonnet-4-6\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7@eu": {
"id": "claude-opus-4-7@eu",
"name": "Claude Opus 4.7 (EU)",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5.5,
"output": 27.5,
"cache_read": 0.55,
"cache_write": 6.875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-opus-4-7@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-max": {
"id": "qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.25,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.7-max\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-it": {
"id": "gemma-4-26b-a4b-it",
"name": "Gemma 4 26B A4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.07,
"output": 0.34,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3@eu": {
"id": "glm-5.3@eu",
"name": "GLM-5.3 (EU)",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.2,
"output": 4.2,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/glm-5.3@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code@eu": {
"id": "kimi-k2.7-code@eu",
"name": "Kimi K2.7 Code (EU)",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.25,
"output": 4.5,
"cache_read": 0.31
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/kimi-k2.7-code@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-2.4T-A95B": {
"id": "qwen3.8-2.4T-A95B",
"name": "Qwen3.8 2.4T A95B",
"description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.8-2.4T-A95B\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-2.4T-A95B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3.5-content-safety": {
"id": "nemotron-3.5-content-safety",
"name": "Nemotron 3.5 Content Safety",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "nemotron",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/nemotron-3.5-content-safety\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3.5-content-safety\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3.7-flash": {
"id": "step-3.7-flash",
"name": "Step 3.7 Flash",
"description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 256000
},
"cost": {
"input": 0.2,
"output": 1.15,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/step-3.7-flash\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"step-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro-0813": {
"id": "deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.32,
"output": 3.96,
"cache_read": 0.044
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-medium-3-5": {
"id": "mistral-medium-3-5",
"name": "mistral-medium-3-5",
"description": "Mistral Medium 3.5 is a dense 128B instruction following model from Mistral AI. It supports text and image inputs with text output, and is designed for agentic workflows, coding, and complex multi step reasoning. It is particularly strong at reliable multi tool calling and long horizon tasks, with a 256K context window, configurable reasoning effort per request, and a custom vision encoder that handles variable image sizes and aspect ratios. Self hostable on as few as four GPUs and available under open weights.",
"family": "mistral-medium",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-04-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.65,
"output": 8.25,
"cache_read": 1.65
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/mistral-medium-3-5\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"mistral-medium-3-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ring-2.6-1t": {
"id": "ring-2.6-1t",
"name": "ring-2.6-1t",
"description": "Inclusion AI ring-2.6-1t",
"family": "ring",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-05-08",
"last_updated": "2026-05-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.3,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/ring-2.6-1t\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"ring-2.6-1t\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-mini@eu": {
"id": "gpt-4.1-mini@eu",
"name": "GPT-4.1 mini (EU)",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.44,
"output": 1.76,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-4.1-mini@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1-mini@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.28,
"output": 0.56,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.7-flash@eu": {
"id": "gemini-3.7-flash@eu",
"name": "Gemini 3.7 Flash (EU)",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.825,
"output": 4.125,
"cache_read": 0.0825
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.7-flash@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.7-flash@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-flash-next": {
"id": "qwen3.8-flash-next",
"name": "Qwen3.8 Flash Next",
"description": "Open-weight experimental preview of the Qwen4 architecture: hybrid-attention MoE (125B total, 6B active) with vision encoder for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-27",
"last_updated": "2026-08-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.8-flash-next\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-flash-next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"leanstral-1-5@eu": {
"id": "leanstral-1-5@eu",
"name": "leanstral-1-5@eu",
"description": "Leanstral 1.5 is an updated Lean 4 formal proof engineering model from Mistral AI, optimized for automated theorem proving and autoformalization. It has 119B total parameters with 6.5B active and supports a 256K token context window. It supports native function calling and structured output.",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-27",
"last_updated": "2026-05-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/leanstral-1-5@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"leanstral-1-5@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro-0813@eu": {
"id": "deepseek-v4-pro-0813@eu",
"name": "DeepSeek V4 Pro 0813 (EU)",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.75,
"output": 3.5,
"cache_read": 0.44
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/deepseek-v4-pro-0813@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-0813@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ling-3.0-tiny": {
"id": "ling-3.0-tiny",
"name": "ling-3.0-tiny",
"description": "Ling-3.0-tiny is an efficient 7.9B parameter MoE model from inclusionAI with only 1.3B active parameters per token. Built for responsive agents, reliable instruction following and multi turn conversation, with a 256K context window, native function calling, prompt caching and switchable Thinking and Instant modes.",
"family": "ling",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/ling-3.0-tiny\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"ling-3.0-tiny\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.3": {
"id": "grok-4.3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"cache_write": 1.25,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"cache_write": 2.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/grok-4.3\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-plus": {
"id": "qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"cache_write": 0.625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.6-plus\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia-nemotron-3-ultra": {
"id": "nvidia-nemotron-3-ultra",
"name": "nvidia-nemotron-3-ultra",
"description": "NVIDIA Nemotron 3 Ultra is NVIDIA's strongest open-weights reasoning model, positioned near GPT-5.4 Mini (xhigh) and ahead of DeepSeek V4-Flash and Qwen3.5-397B-A17B.",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": false,
"structured_output": false,
"release_date": "2026-06-23",
"last_updated": "2026-06-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.5,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/nvidia-nemotron-3-ultra\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"nvidia-nemotron-3-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia-nemotron-3-super-120b-a12b": {
"id": "nvidia-nemotron-3-super-120b-a12b",
"name": "nvidia-nemotron-3-super-120b-a12b",
"description": "NVIDIA Nemotron 3 Super is a hybrid Mixture-of-Experts (MoE) model engineered for highest compute efficiency and accuracy in multi-agent applications and specialized agentic systems. It is optimized to run many collaborating agents per application on a single GPU, delivering high accuracy for reasoning, tool use, and instruction following.",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.1,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/nvidia-nemotron-3-super-120b-a12b\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"nvidia-nemotron-3-super-120b-a12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.7-highspeed": {
"id": "minimax-m2.7-highspeed",
"name": "MiniMax-M2.7-highspeed",
"description": "Low-latency M2.7 variant for interactive coding plans and agent loops",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.06,
"cache_write": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/minimax-m2.7-highspeed\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-sol": {
"id": "gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 4,
"output": 20,
"cache_read": 0.4,
"cache_write": 5,
"tiers": [
{
"input": 8,
"output": 30,
"cache_read": 0.8,
"cache_write": 10,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 8,
"output": 30,
"cache_read": 0.8,
"cache_write": 10
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.6-sol\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fugu-ultra": {
"id": "fugu-ultra",
"name": "Fugu Ultra",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/fugu-ultra\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"fugu-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5.1@eu": {
"id": "claude-fable-5.1@eu",
"name": "Claude Fable 5.1 (EU)",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 11,
"output": 55,
"cache_read": 0.275,
"cache_write": 13.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-fable-5.1@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5.1@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-pro-image": {
"id": "gemini-3-pro-image",
"name": "Nano Banana Pro",
"description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 32768
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 4.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3-pro-image\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-pro-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-27b": {
"id": "qwen3.5-27b",
"name": "Qwen3.5 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.26,
"output": 2.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.5-27b\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6@eu": {
"id": "claude-opus-4-6@eu",
"name": "Claude Opus 4.6 (EU)",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5.5,
"output": 27.5,
"cache_read": 0.55,
"cache_write": 6.875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-opus-4-6@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-35b-a3b": {
"id": "qwen3.5-35b-a3b",
"name": "Qwen3.5 35B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.14,
"output": 1,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.5-35b-a3b\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5": {
"id": "claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-opus-5\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.7": {
"id": "minimax-m2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06,
"cache_write": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/minimax-m2.7\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/kimi-k2.6\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro-preview": {
"id": "gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 4.5,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 9,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 9
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kat-coder-pro": {
"id": "kat-coder-pro",
"name": "kat-coder-pro",
"description": "KAT-Coder-Pro V2 by KwaiKAT is a non-reasoning model optimized for agentic coding. It delivers strong performance on reasoning-style tasks while requiring significantly fewer output tokens than peer models. With the 1210 release, it achieved a score of 64 on the Artificial Analysis Intelligence Index, placing it in the global Top 10 and ranking first among all non-reasoning models.",
"family": "kat-coder",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2026-03-27",
"last_updated": "2026-03-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/kat-coder-pro\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"kat-coder-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-terra@eu": {
"id": "gpt-5.6-terra@eu",
"name": "GPT-5.6 Terra (EU)",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.2,
"output": 13.2,
"cache_read": 0.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.6-terra@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-terra@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.8,
"output": 2.55,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/glm-5.2\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4@eu": {
"id": "gpt-5.4@eu",
"name": "GPT-5.4 (EU)",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.4@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-6-astra": {
"id": "gpt-6-astra",
"name": "GPT-6 Astra",
"description": "GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.",
"family": "gpt-astra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-04-30",
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"tiers": [
{
"input": 20,
"output": 75,
"cache_read": 2,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 20,
"output": 75,
"cache_read": 2
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-6-astra\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-6-astra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.8-flash@eu": {
"id": "gemini-3.8-flash@eu",
"name": "Gemini 3.8 Flash (EU)",
"description": "Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.825,
"output": 4.125,
"cache_read": 0.0825
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.8-flash@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.8-flash@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5": {
"id": "claude-opus-4-5",
"name": "Claude Opus 4.5 (latest)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-opus-4-5\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-nano@eu": {
"id": "gpt-5-nano@eu",
"name": "GPT-5 Nano (EU)",
"description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 0.055,
"output": 0.44,
"cache_read": 0.0055
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5-nano@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-nano@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m3": {
"id": "minimax-m3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/minimax-m3\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.28,
"output": 0.56,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/deepseek-v4-flash\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code": {
"id": "kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/kimi-k2.7-code\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5@eu": {
"id": "claude-sonnet-5@eu",
"name": "Claude Sonnet 5 (EU)",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2.2,
"output": 11,
"cache_read": 0.22,
"cache_write": 2.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-sonnet-5@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-5@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.5": {
"id": "grok-4.5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5,
"cache_write": 2,
"tiers": [
{
"input": 4,
"output": 12,
"cache_read": 1,
"cache_write": 4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 12,
"cache_read": 1,
"cache_write": 4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/grok-4.5\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.6-flash": {
"id": "gemini-3.6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 1.5,
"output": 7,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.6-flash\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-nano-omni-30b-a3b-reasoning": {
"id": "nemotron-3-nano-omni-30b-a3b-reasoning",
"name": "Nemotron 3 Nano Omni 30B A3B Reasoning",
"description": "Open Nemotron omni model combining reasoning with text, vision, and audio",
"family": "nemotron",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-28",
"last_updated": "2026-04-28",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 20480
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/nemotron-3-nano-omni-30b-a3b-reasoning\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-nano-omni-30b-a3b-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5@eu": {
"id": "gpt-5.5@eu",
"name": "GPT-5.5 (EU)",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.5@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.75,
"output": 16.5,
"cache_read": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.4\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite": {
"id": "gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-1.8": {
"id": "seed-1.8",
"name": "seed-1.8",
"description": "Optimized specifically for multimodal agent scenarios. It features enhanced agent capabilities, upgraded multimodal comprehension, and more flexible context management.",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-05-27",
"last_updated": "2026-05-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/seed-1.8\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"seed-1.8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-build-0.1": {
"id": "grok-build-0.1",
"name": "Grok Build 0.1",
"description": "Fast Grok coding model tuned for agentic engineering and iterative edits",
"family": "grok-build",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 1,
"output": 2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/grok-build-0.1\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"grok-build-0.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-mini@eu": {
"id": "gpt-5-mini@eu",
"name": "GPT-5 Mini (EU)",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 0.275,
"output": 2.2,
"cache_read": 0.0275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5-mini@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-mini@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-nano@eu": {
"id": "gpt-4.1-nano@eu",
"name": "GPT-4.1 nano (EU)",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.11,
"output": 0.44,
"cache_read": 0.0275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-4.1-nano@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1-nano@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4.1-flash": {
"id": "deepseek-v4.1-flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393216
},
"cost": {
"input": 0.22,
"output": 0.66,
"cache_read": 0.007
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/deepseek-v4.1-flash\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4.1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"hy3": {
"id": "hy3",
"name": "Hy3",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.14,
"output": 0.58,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/hy3\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5@eu": {
"id": "claude-fable-5@eu",
"name": "Claude Fable 5 (EU)",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 11,
"output": 55,
"cache_read": 1.1,
"cache_write": 13.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-fable-5@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ling-2.6-1t": {
"id": "ling-2.6-1t",
"name": "ling-2.6-1t",
"description": "Inclusion AI ling-2.6-1t",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.3,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/ling-2.6-1t\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"ling-2.6-1t\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-medium-latest": {
"id": "mistral-medium-latest",
"name": "Mistral Medium (latest)",
"description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
"family": "mistral-medium",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-29",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.44,
"output": 2.2,
"cache_read": 0.44
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/mistral-medium-latest\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"mistral-medium-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-2.0-pro": {
"id": "seed-2.0-pro",
"name": "Seed 2.0 Pro",
"description": "Flagship ByteDance Seed 2.0 model for complex multimodal reasoning and long-horizon agent workflows",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/seed-2.0-pro\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"seed-2.0-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1@eu": {
"id": "glm-5.1@eu",
"name": "GLM-5.1 (EU)",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 200000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 1.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/glm-5.1@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-flash-next@eu": {
"id": "qwen3.8-flash-next@eu",
"name": "Qwen3.8 Flash Next (EU)",
"description": "Open-weight experimental preview of the Qwen4 architecture: hybrid-attention MoE (125B total, 6B active) with vision encoder for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-27",
"last_updated": "2026-08-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.8-flash-next@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-flash-next@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingcap-qwen3.6-27b": {
"id": "thinkingcap-qwen3.6-27b",
"name": "thinkingcap-qwen3.6-27b",
"description": "ThinkingCap-Qwen3.6-27B is a reasoning tuned model from BottlecapAI built on Qwen3.6 27B. It supports extended thinking with tool calling and a 256K context window. Served via Sference.",
"family": "qwen3.6",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-13",
"last_updated": "2026-07-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 2.6,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/thinkingcap-qwen3.6-27b\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"thinkingcap-qwen3.6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-ultra-nvfp4": {
"id": "nemotron-3-ultra-nvfp4",
"name": "nemotron-3-ultra-nvfp4",
"description": "Nemotron-3-Ultra-550B-A55B-NVFP4 is a frontier-scale large language model (LLM) trained by NVIDIA, designed to deliver strong agentic, reasoning, and conversational capabilities. It is optimized for the most demanding workloads, including complex multi-step agents, long-context analysis, and high-accuracy reasoning over code, math, and science. The model employs a hybrid Latent Mixture-of-Experts (LatentMoE) architecture, utilizing interleaved Mamba-2 and MoE layers, along with select Attention layers. Like the Super model, the Ultra model incorporates Multi-Token Prediction (MTP) layers for faster text generation and improved quality, and it is trained using an NVFP4 pre-training recipe to maximize compute efficiency. The model has 55B active parameters and 550B parameters in total.",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/nemotron-3-ultra-nvfp4\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-ultra-nvfp4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m3@eu": {
"id": "minimax-m3@eu",
"name": "MiniMax-M3 (EU)",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.4,
"output": 2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/minimax-m3@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m3@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-opus-4-6\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"devstral-latest@eu": {
"id": "devstral-latest@eu",
"name": "devstral-latest@eu",
"description": "An enterprise grade text model, that excels at using tools to explore codebases, editing multiple files and power software engineering agents.",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-27",
"last_updated": "2026-05-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.44,
"output": 2.2,
"cache_read": 0.44
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/devstral-latest@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"devstral-latest@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash": {
"id": "gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"cache_write": 1.583
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.5-flash\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-medium-3-5@eu": {
"id": "mistral-medium-3-5@eu",
"name": "mistral-medium-3-5@eu",
"description": "Mistral Medium 3.5 is a dense 128B instruction following model from Mistral AI. It supports text and image inputs with text output, and is designed for agentic workflows, coding, and complex multi step reasoning. It is particularly strong at reliable multi tool calling and long horizon tasks, with a 256K context window, configurable reasoning effort per request, and a custom vision encoder that handles variable image sizes and aspect ratios. Self hostable on as few as four GPUs and available under open weights.",
"family": "mistral-medium",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-04-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.65,
"output": 8.25,
"cache_read": 1.65
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/mistral-medium-3-5@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"mistral-medium-3-5@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-luna": {
"id": "gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"tiers": [
{
"input": 0.4,
"output": 1.8,
"cache_read": 0.04,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.04
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.6-luna\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-lightning-3.5-30b-a3b": {
"id": "nemotron-lightning-3.5-30b-a3b",
"name": "nemotron-lightning-3.5-30b-a3b",
"description": "Nemotron-Lightning-3.5-30B-A3B is a 30B-parameter Mixture-of-Experts language model (3B active) from NVIDIA's Nemotron-H family, built on a hybrid Mamba-Transformer architecture for efficient long-context inference. Like other models in the family, it responds to queries by first generating a reasoning trace and then concluding with a final response, with reasoning behavior configurable through a flag in the chat template. It includes a multi-token prediction (MTP) speculative decoding head for low-latency serving.",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-15",
"last_updated": "2026-08-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.05,
"output": 0.2,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/nemotron-lightning-3.5-30b-a3b\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-lightning-3.5-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-2.0-code": {
"id": "seed-2.0-code",
"name": "Seed 2.0 Code",
"description": "ByteDance Seed coding model for multimodal software engineering and long-running agents",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/seed-2.0-code\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"seed-2.0-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7": {
"id": "claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-opus-4-7\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/kimi-k3\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-medium-latest@eu": {
"id": "mistral-medium-latest@eu",
"name": "Mistral Medium (latest) (EU)",
"description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
"family": "mistral-medium",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-29",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.44,
"output": 2.2,
"cache_read": 0.44
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/mistral-medium-latest@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"mistral-medium-latest@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-sol@eu": {
"id": "gpt-5.6-sol@eu",
"name": "GPT-5.6 Sol (EU)",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 4.4,
"output": 22,
"cache_read": 0.44
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.6-sol@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-sol@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-fast": {
"id": "glm-5.2-fast",
"name": "glm-5.2-fast",
"description": "GLM-5.2 introduces a robust 1M-token context and advanced, multi-effort coding capabilities to significantly enhance performance on long-horizon tasks. Its new IndexShare architecture and improved MTP layer simultaneously boost efficiency by reducing per-token FLOPs and increasing speculative decoding lengths. A 743B-parameter model in Zhipu AI's GLM series, designed to plan, execute, and iterate autonomously on extended, engineering-grade tasks.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-13",
"last_updated": "2026-07-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2.1,
"output": 6.6,
"cache_read": 0.21
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/glm-5.2-fast\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.3-codex": {
"id": "gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.3-codex\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-image": {
"id": "gemini-3.1-flash-image",
"name": "Nano Banana 2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.1-flash-image\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 262144
},
"cost": {
"input": 0.2,
"output": 0.6,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/glm-5.3-flash\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-6@eu": {
"id": "claude-sonnet-4-6@eu",
"name": "Claude Sonnet 4.6 (EU)",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3.3,
"output": 16.5,
"cache_read": 0.3,
"cache_write": 4.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-sonnet-4-6@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-fable-5\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash-lite": {
"id": "gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"laguna-m.1": {
"id": "laguna-m.1",
"name": "Laguna M.1",
"description": "Poolside's open-weight model for agentic coding and long-horizon work",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-28",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/laguna-m.1\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"laguna-m.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"leanstral-1-5": {
"id": "leanstral-1-5",
"name": "leanstral-1-5",
"description": "Leanstral 1.5 is an updated Lean 4 formal proof engineering model from Mistral AI, optimized for automated theorem proving and autoformalization. It has 119B total parameters with 6.5B active and supports a 256K token context window. It supports native function calling and structured output.",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-27",
"last_updated": "2026-05-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/leanstral-1-5\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"leanstral-1-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash@eu": {
"id": "glm-5.3-flash@eu",
"name": "GLM-5.3-Flash (EU)",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 262144
},
"cost": {
"input": 0.2,
"output": 0.6,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/glm-5.3-flash@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-nano": {
"id": "gpt-5.4-nano",
"name": "GPT-5.4 nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.4-nano\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"laguna-xs.2": {
"id": "laguna-xs.2",
"name": "Laguna XS.2",
"description": "Agentic coding model from Poolside in the XS size class for local deployment",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-28",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/laguna-xs.2\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"laguna-xs.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5-pro": {
"id": "gpt-5.5-pro",
"name": "GPT-5.5 Pro",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.5-pro\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-nano-omni": {
"id": "nemotron-3-nano-omni",
"name": "nemotron-3-nano-omni",
"description": "The most open, efficient, and accurate omni modal reasoning model for agentic AI.",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-20",
"last_updated": "2026-05-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 300000,
"output": 300000
},
"cost": {
"input": 0.06,
"output": 0.24,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/nemotron-3-nano-omni\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-nano-omni\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-super-120b-a12b": {
"id": "nemotron-3-super-120b-a12b",
"name": "Nemotron 3 Super 120B A12B",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/nemotron-3-super-120b-a12b\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-super-120b-a12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5@eu": {
"id": "gpt-5@eu",
"name": "GPT-5 (EU)",
"description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.375,
"output": 11,
"cache_read": 0.1375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingcap-qwen3.6-27b@eu": {
"id": "thinkingcap-qwen3.6-27b@eu",
"name": "thinkingcap-qwen3.6-27b@eu",
"description": "ThinkingCap-Qwen3.6-27B is a reasoning tuned model from BottlecapAI built on Qwen3.6 27B. It supports extended thinking with tool calling and a 256K context window. Served via Sference.",
"family": "qwen3.6",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-13",
"last_updated": "2026-07-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 2.6,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/thinkingcap-qwen3.6-27b@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"thinkingcap-qwen3.6-27b@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-2.0-mini": {
"id": "seed-2.0-mini",
"name": "Seed 2.0 Mini",
"description": "Lightweight ByteDance Seed 2.0 model for low-latency multimodal reasoning and high-volume tasks",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/seed-2.0-mini\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"seed-2.0-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6@eu": {
"id": "kimi-k2.6@eu",
"name": "Kimi K2.6 (EU)",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/kimi-k2.6@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-glimmer-30b": {
"id": "muse-glimmer-30b",
"name": "Muse Glimmer 30B",
"description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2026-01-04",
"release_date": "2026-08-10",
"last_updated": "2026-08-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 20480
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/muse-glimmer-30b\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"muse-glimmer-30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-nano-omni@eu": {
"id": "nemotron-3-nano-omni@eu",
"name": "nemotron-3-nano-omni@eu",
"description": "The most open, efficient, and accurate omni modal reasoning model for agentic AI.",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-20",
"last_updated": "2026-05-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 300000,
"output": 300000
},
"cost": {
"input": 0.06,
"output": 0.24,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/nemotron-3-nano-omni@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-nano-omni@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ling-2.6-flash": {
"id": "ling-2.6-flash",
"name": "ling-2.6-flash",
"description": "Inclusion AI ling-2.6-flash",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.1,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/ling-2.6-flash\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"ling-2.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro@eu": {
"id": "deepseek-v4-pro@eu",
"name": "DeepSeek V4 Pro (EU)",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.75,
"output": 3.5,
"cache_read": 0.44
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/deepseek-v4-pro@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-flash": {
"id": "qwen3.8-flash",
"name": "Qwen3.8 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.16,
"output": 0.47,
"cache_read": 0.016,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.8-flash\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4@eu": {
"id": "claude-sonnet-4@eu",
"name": "Claude Sonnet 4 (latest) (EU)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75,
"tiers": [
{
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-sonnet-4@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash-lite@eu": {
"id": "gemini-3.5-flash-lite@eu",
"name": "Gemini 3.5 Flash Lite (EU)",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.33,
"output": 2.75,
"cache_read": 0.033
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.5-flash-lite@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash-lite@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-mini": {
"id": "gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.4-mini\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inkling": {
"id": "inkling",
"name": "Inkling",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 32768
},
"cost": {
"input": 1.87,
"output": 4.68,
"cache_read": 0.374
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/inkling\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5@eu": {
"id": "claude-opus-5@eu",
"name": "Claude Opus 5 (EU)",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5.5,
"output": 27.5,
"cache_read": 0.55,
"cache_write": 6.875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-opus-5@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-5@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2@eu": {
"id": "glm-5.2@eu",
"name": "GLM-5.2 (EU)",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4.2,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/glm-5.2@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it": {
"id": "gemma-4-31b-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemma-4-31b-it\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.6": {
"id": "grok-4.6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5,
"cache_write": 2,
"tiers": [
{
"input": 4,
"output": 12,
"cache_read": 1,
"cache_write": 4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 12,
"cache_read": 1,
"cache_write": 4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/grok-4.6\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5@eu": {
"id": "claude-haiku-4-5@eu",
"name": "Claude Haiku 4.5 (latest) (EU)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1.1,
"output": 5.5,
"cache_read": 0.11,
"cache_write": 1.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-haiku-4-5@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o-mini@eu": {
"id": "gpt-4o-mini@eu",
"name": "GPT-4o mini (EU)",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16000
},
"cost": {
"input": 0.165,
"output": 0.66,
"cache_read": 0.0825
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-4o-mini@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4o-mini@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5": {
"id": "claude-haiku-4-5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-haiku-4-5\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5": {
"id": "claude-sonnet-4-5",
"name": "Claude Sonnet 4.5 (latest)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75,
"tiers": [
{
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-sonnet-4-5\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1": {
"id": "claude-opus-4-1",
"name": "Claude Opus 4.1 (latest)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-opus-4-1\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max": {
"id": "qwen3.8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.8-max\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/glm-5.1\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"devstral-latest": {
"id": "devstral-latest",
"name": "devstral-latest",
"description": "An enterprise grade text model, that excels at using tools to explore codebases, editing multiple files and power software engineering agents.",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-05-27",
"last_updated": "2026-05-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.44,
"output": 2.2,
"cache_read": 0.44
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/devstral-latest\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"devstral-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3@eu": {
"id": "kimi-k3@eu",
"name": "Kimi K3 (EU)",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/kimi-k3@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-plus": {
"id": "qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.32,
"output": 1.28,
"cache_read": 0.032,
"cache_write": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.7-plus\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite@eu": {
"id": "gemini-2.5-flash-lite@eu",
"name": "Gemini 2.5 Flash-Lite (EU)",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01,
"cache_write": 0.18333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-2.5-flash-lite@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-2603": {
"id": "mistral-small-2603",
"name": "Mistral Small 4",
"description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents",
"family": "mistral-small",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.165,
"output": 0.66,
"cache_read": 0.165
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/mistral-small-2603\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small-2603\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.8-flash": {
"id": "gemini-3.8-flash",
"name": "Gemini 3.8 Flash",
"description": "Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.8-flash\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-opus-4-8\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.2-beta": {
"id": "grok-4.2-beta",
"name": "grok-4.2-beta",
"description": "Grok 4.20 Beta is xAI's newest flagship model with industry-leading speed and agentic tool calling capabilities. It combines the lowest hallucination rate on the market with strict prompt adherance, delivering consistently precise and truthful responses.",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-19",
"last_updated": "2026-03-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2,
"tiers": [
{
"input": 4,
"output": 12,
"cache_read": 0.4,
"cache_write": 4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 12,
"cache_read": 0.4,
"cache_write": 4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/grok-4.2-beta\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4.2-beta\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.32,
"output": 3.96,
"cache_read": 0.044
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/deepseek-v4-pro\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5@eu": {
"id": "claude-sonnet-4-5@eu",
"name": "Claude Sonnet 4.5 (latest) (EU)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3.3,
"output": 16.5,
"cache_read": 0.3,
"cache_write": 4.125,
"tiers": [
{
"input": 6.6,
"output": 24.75,
"cache_read": 0.6,
"cache_write": 8.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 6.6,
"output": 24.75,
"cache_read": 0.6,
"cache_write": 8.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-sonnet-4-5@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-2603@eu": {
"id": "mistral-small-2603@eu",
"name": "Mistral Small 4 (EU)",
"description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents",
"family": "mistral-small",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.165,
"output": 0.66,
"cache_read": 0.165
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/mistral-small-2603@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small-2603@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-2b": {
"id": "qwen3.5-2b",
"name": "qwen3.5-2b",
"description": "Qwen3.5-2B is a compact yet capable model from Alibaba's Qwen3.5 series. It features a 262K token context window, support for 201 languages, thinking/reasoning mode, and tool calling for agentic workflows. A strong choice for prototyping, fine-tuning, and efficient multilingual deployments.",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-10",
"last_updated": "2026-03-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.02,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.5-2b\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-2b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5@eu": {
"id": "claude-opus-4-5@eu",
"name": "Claude Opus 4.5 (latest) (EU)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5.5,
"output": 27.5,
"cache_read": 0.55,
"cache_write": 6.875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-opus-4-5@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-5@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-pro": {
"id": "gpt-5.4-pro",
"name": "GPT-5.4 Pro",
"description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180,
"cache_read": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.4-pro\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite@eu": {
"id": "gemini-3.1-flash-lite@eu",
"name": "Gemini 3.1 Flash Lite (EU)",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.275,
"output": 1.65,
"cache_read": 0.0275,
"cache_write": 0.091663
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.1-flash-lite@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.7-flash": {
"id": "gemini-3.7-flash",
"name": "Gemini 3.7 Flash",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.7-flash\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash@eu": {
"id": "gemini-2.5-flash@eu",
"name": "Gemini 2.5 Flash (EU)",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.075,
"cache_write": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-2.5-flash@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro@eu": {
"id": "gemini-2.5-pro@eu",
"name": "Gemini 2.5 Pro (EU)",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.31,
"cache_write": 2.375,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.62,
"cache_write": 4.75,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.62,
"cache_write": 4.75
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-2.5-pro@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-terra": {
"id": "gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.6-terra\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.2,
"output": 4.2,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/glm-5.3\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-ultra-550b-a55b": {
"id": "nemotron-3-ultra-550b-a55b",
"name": "Nemotron 3 Ultra 550B A55B",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/nemotron-3-ultra-550b-a55b\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-ultra-550b-a55b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3.5-lightning-30b-a3b": {
"id": "nemotron-3.5-lightning-30b-a3b",
"name": "nemotron-3.5-lightning-30b-a3b",
"description": "NVIDIA Nemotron 3.5 Lightning 30B-A3B is a hybrid Mamba-2 + MoE + Attention model with 30B total and 3B active parameters, pre-trained on over 20T tokens with an NVFP4 recipe and Multi-Token Prediction for fast generation. Up to 1M token context for long-running autonomous agents, sub-agent workhorse deployments, and agentic workflows. Supports reasoning and tool calling. English and coding languages plus Spanish, French, German, Italian, and Japanese. Open weights under the OpenMDW License Agreement v1.1. Part of the NVIDIA Nemotron family.",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-08-11",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/nemotron-3.5-lightning-30b-a3b\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3.5-lightning-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inkling-256k": {
"id": "inkling-256k",
"name": "inkling-256k",
"description": "Inkling 256K is the extended context variant of Inkling, a large MoE hybrid reasoning model from Thinking Machines with audio and vision input support and a 256K context window.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 1.87,
"output": 4.68,
"cache_read": 0.374
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/inkling-256k\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"inkling-256k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash@eu": {
"id": "gemini-3.5-flash@eu",
"name": "Gemini 3.5 Flash (EU)",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 1.65,
"output": 9.9,
"cache_read": 0.165,
"cache_write": 1.7413
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gemini-3.5-flash@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-luna@eu": {
"id": "gpt-5.6-luna@eu",
"name": "GPT-5.6 Luna (EU)",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.22,
"output": 1.32,
"cache_read": 0.022
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.6-luna@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-luna@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8@eu": {
"id": "claude-opus-4-8@eu",
"name": "Claude Opus 4.8 (EU)",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5.5,
"output": 27.5,
"cache_read": 0.55,
"cache_write": 6.875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-opus-4-8@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o4-mini@eu": {
"id": "o4-mini@eu",
"name": "o4-mini (EU)",
"description": "Fast o-series model for compact reasoning, coding, and tool use",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.21,
"output": 4.84,
"cache_read": 0.3025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/o4-mini@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"o4-mini@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1@eu": {
"id": "gpt-4.1@eu",
"name": "GPT-4.1 (EU)",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 2.2,
"output": 8.8,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-4.1@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5": {
"id": "claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-sonnet-5\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731@eu": {
"id": "deepseek-v4-flash-0731@eu",
"name": "DeepSeek V4 Flash 0731 (EU)",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.28,
"output": 0.56,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/deepseek-v4-flash-0731@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5.1": {
"id": "claude-fable-5.1",
"name": "Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/claude-fable-5.1\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-2.4T-A95B@eu": {
"id": "qwen3.8-2.4T-A95B@eu",
"name": "Qwen3.8 2.4T A95B (EU)",
"description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 262144
},
"cost": {
"input": 2.5,
"output": 6,
"cache_read": 0.63
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/qwen3.8-2.4T-A95B@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-2.4T-A95B@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5": {
"id": "mimo-v2.5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/mimo-v2.5\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1@eu": {
"id": "gpt-5.1@eu",
"name": "GPT-5.1 (EU)",
"description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.375,
"output": 11,
"cache_read": 0.1375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.1@eu\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1@eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-pro": {
"id": "mimo-v2.5-pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.0036
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/mimo-v2.5-pro\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5.5,
"output": 33,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"requesty/gpt-5.5\", apiKey: processEnvironment[\"REQUESTY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.requesty.ai/v1\")!,\n apiKey: processEnvironment[\"REQUESTY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"llmtr": {
"id": "llmtr",
"name": "LLMTR",
"baseURL": "https://llmtr.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"LLMTR_API_KEY"
],
"doc": "https://llmtr.com/docs",
"modelCount": 32,
"models": {
"medgemma-4b": {
"id": "medgemma-4b",
"name": "MedGemma 4B",
"description": "Multimodal medical-domain Gemma variant for text and image analysis",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-04-26",
"last_updated": "2026-08-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"status": "deprecated",
"cost": {
"input": 3,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/medgemma-4b\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"medgemma-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-glimmer-30b-tr": {
"id": "muse-glimmer-30b-tr",
"name": "Muse Glimmer 30B (TR)",
"description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-01-04",
"release_date": "2026-08-10",
"last_updated": "2026-08-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 2,
"output": 5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/muse-glimmer-30b-tr\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"muse-glimmer-30b-tr\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4": {
"id": "gemma-4",
"name": "Gemma 4",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 2,
"output": 5,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/gemma-4\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"magibu-11b-v8": {
"id": "magibu-11b-v8",
"name": "Magibu 11B v8",
"description": "Turkish-language chat model for instruction following and assistant flows",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-06-05",
"last_updated": "2026-08-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 0.1,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/magibu-11b-v8\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"magibu-11b-v8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-6-35b": {
"id": "qwen3-6-35b",
"name": "Qwen3.6 35B-A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16384,
"output": 16384
},
"cost": {
"input": 5,
"output": 10
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen3-6-35b\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-6-35b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"trendyol-asure-12b": {
"id": "trendyol-asure-12b",
"name": "Trendyol Asure 12B",
"description": "Turkish-language multimodal instruct model built on Gemma 3 12B for e-commerce text, chat, and image-text tasks",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-02-19",
"last_updated": "2026-02-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 40960
},
"cost": {
"input": 0.1,
"output": 0.5,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/trendyol-asure-12b\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"trendyol-asure-12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-plus": {
"id": "qwen/qwen3-coder-plus",
"name": "Qwen3 Coder Plus",
"description": "Hosted Qwen coder for software agents, repo edits, and long-context code",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 1,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen/qwen3-coder-plus\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-flash": {
"id": "qwen/qwen3-coder-flash",
"name": "Qwen3 Coder Flash",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen/qwen3-coder-flash\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-plus": {
"id": "qwen/qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen/qwen3.6-plus\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-flash": {
"id": "qwen/qwen-flash",
"name": "Qwen Flash",
"description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen/qwen-flash\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-plus": {
"id": "qwen/qwen3-vl-plus",
"name": "Qwen3-VL Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 1.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen/qwen3-vl-plus\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-397b-a17b": {
"id": "qwen/qwen3.5-397b-a17b",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-max": {
"id": "qwen/qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 1.2,
"output": 6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen/qwen3-max\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-plus": {
"id": "qwen/qwen-plus",
"name": "Qwen Plus",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01-25",
"last_updated": "2025-09-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen/qwen-plus\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-flash": {
"id": "qwen/qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen/qwen3.6-flash\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.7-plus": {
"id": "qwen/qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.4,
"output": 1.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen/qwen3.7-plus\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-plus": {
"id": "qwen/qwen3.5-plus",
"name": "Qwen3.5 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/qwen/qwen3.5-plus\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-xs-2.1": {
"id": "poolside/laguna-xs-2.1",
"name": "Laguna XS 2.1",
"description": "Agentic coding model from Poolside in the XS size class for local deployment",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-02",
"last_updated": "2026-07-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/poolside/laguna-xs-2.1\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-xs-2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo/mimo-v2.5": {
"id": "mimo/mimo-v2.5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/mimo/mimo-v2.5\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"mimo/mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo/mimo-v2.5-pro": {
"id": "mimo/mimo-v2.5-pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.435,
"output": 0.87
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/mimo/mimo-v2.5-pro\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"mimo/mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash-lite": {
"id": "google/gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash-Lite",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/google/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/inkling-small": {
"id": "thinkingmachines/inkling-small",
"name": "Inkling Small",
"description": "Multimodal MoE reasoning model (276B total, 12B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-30",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.58,
"output": 1.44
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/thinkingmachines/inkling-small\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/inkling-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/inkling": {
"id": "thinkingmachines/inkling",
"name": "Inkling",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.87,
"output": 4.68
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/thinkingmachines/inkling\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.2-contributor": {
"id": "meta/muse-spark-1.2-contributor",
"name": "Muse Spark 1.2 Contributor",
"description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/meta/muse-spark-1.2-contributor\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.2-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"publicai/apertus-8b-instruct": {
"id": "publicai/apertus-8b-instruct",
"name": "Apertus 8B Instruct",
"description": "Fully open 8B multilingual LLM supporting 1800+ languages with 65K context. Trained on compliant open data. Apache 2.0, EU AI Act compliant.",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-02",
"last_updated": "2025-09-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 8192
},
"cost": {
"input": 0.1,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/publicai/apertus-8b-instruct\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"publicai/apertus-8b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"publicai/apertus-70b-instruct": {
"id": "publicai/apertus-70b-instruct",
"name": "Apertus 70B Instruct",
"description": "Fully open 70B multilingual LLM supporting 1800+ languages with 65K context. Trained on 15T tokens of compliant open data. Apache 2.0, EU AI Act compliant.",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-02",
"last_updated": "2025-09-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 8192
},
"cost": {
"input": 0.82,
"output": 2.92
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/publicai/apertus-70b-instruct\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"publicai/apertus-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sakana/fugu-ultra": {
"id": "sakana/fugu-ultra",
"name": "Fugu Ultra",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 5,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/sakana/fugu-ultra\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"sakana/fugu-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"upstage/solar-pro4": {
"id": "upstage/solar-pro4",
"name": "Solar Pro 4",
"description": "Upstage's flagship model, specialized for agentic use",
"family": "solar-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02",
"release_date": "2026-08-06",
"last_updated": "2026-08-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 524288,
"output": 131072
},
"cost": {
"input": 0.03,
"output": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/upstage/solar-pro4\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"upstage/solar-pro4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"upstage/solar-pro3": {
"id": "upstage/solar-pro3",
"name": "Solar Pro 3",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "solar-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/upstage/solar-pro3\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"upstage/solar-pro3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"upstage/solar-pro2": {
"id": "upstage/solar-pro2",
"name": "Solar Pro 2",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "solar-pro",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-05-20",
"last_updated": "2025-05-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 8192
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/upstage/solar-pro2\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"upstage/solar-pro2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/voxtral-small-latest": {
"id": "mistral/voxtral-small-latest",
"name": "Voxtral Small (latest)",
"description": "Instruct model with native audio input for speech understanding and tool use",
"family": "voxtral",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-07-15",
"last_updated": "2025-07-15",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 32000
},
"cost": {
"input": 0.1,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/mistral/voxtral-small-latest\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"mistral/voxtral-small-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/sonar-deep-research": {
"id": "perplexity/sonar-deep-research",
"name": "Sonar Deep Research",
"description": "Sonar search model for autonomous research and citation-backed long-form reports",
"family": "sonar",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2025-02-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"reasoning": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmtr/perplexity/sonar-deep-research\", apiKey: processEnvironment[\"LLMTR_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://llmtr.com/v1\")!,\n apiKey: processEnvironment[\"LLMTR_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/sonar-deep-research\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"xiaomi": {
"id": "xiaomi",
"name": "Xiaomi",
"baseURL": "https://api.xiaomimimo.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"XIAOMI_API_KEY"
],
"doc": "https://platform.xiaomimimo.com/#/docs",
"modelCount": 6,
"models": {
"mimo-v2.5-pro-ultraspeed": {
"id": "mimo-v2.5-pro-ultraspeed",
"name": "MiMo-V2.5-Pro-UltraSpeed",
"description": "MiMo pro model for strong multimodal reasoning and agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-06-08",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"status": "beta",
"cost": {
"input": 1.305,
"output": 2.61,
"cache_read": 0.0108
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi/mimo-v2.5-pro-ultraspeed\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-pro-ultraspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-flash": {
"id": "mimo-v2-flash",
"name": "MiMo-V2-Flash",
"description": "Legacy model retained for compatibility with older integrations",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12-01",
"release_date": "2025-12-16",
"last_updated": "2026-06-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"status": "deprecated",
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi/mimo-v2-flash\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-pro": {
"id": "mimo-v2-pro",
"name": "MiMo-V2-Pro",
"description": "Earlier MiMo Pro model for multimodal agents, reasoning, and code tasks",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-03-18",
"last_updated": "2026-06-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"status": "deprecated",
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.0036
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi/mimo-v2-pro\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-omni": {
"id": "mimo-v2-omni",
"name": "MiMo-V2-Omni",
"description": "Legacy model retained for compatibility with older integrations",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-03-18",
"last_updated": "2026-06-24",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 131072
},
"status": "deprecated",
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi/mimo-v2-omni\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-omni\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5": {
"id": "mimo-v2.5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-06-24",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi/mimo-v2.5\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-pro": {
"id": "mimo-v2.5-pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-06-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.0036
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xiaomi/mimo-v2.5-pro\", apiKey: processEnvironment[\"XIAOMI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.xiaomimimo.com/v1\")!,\n apiKey: processEnvironment[\"XIAOMI_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"huggingface": {
"id": "huggingface",
"name": "Hugging Face",
"baseURL": "https://router.huggingface.co/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"HF_TOKEN"
],
"doc": "https://huggingface.co/docs/inference-providers",
"modelCount": 77,
"models": {
"stepfun-ai/Step-3.7-Flash": {
"id": "stepfun-ai/Step-3.7-Flash",
"name": "Step 3.7 Flash",
"description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 256000
},
"cost": {
"input": 0.2,
"output": 1.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/stepfun-ai/Step-3.7-Flash\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"stepfun-ai/Step-3.7-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun-ai/Step-3.5-Flash": {
"id": "stepfun-ai/Step-3.5-Flash",
"name": "Step 3.5 Flash",
"description": "StepFun flash lane for quick multimodal reasoning and coding assistance",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-29",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 256000
},
"cost": {
"input": 0.1,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/stepfun-ai/Step-3.5-Flash\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"stepfun-ai/Step-3.5-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3": {
"id": "deepseek-ai/DeepSeek-V3",
"name": "DeepSeek-V3",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-12-26",
"last_updated": "2024-12-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 64000,
"output": 8192
},
"cost": {
"input": 0.4,
"output": 1.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-V3\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Flash": {
"id": "deepseek-ai/DeepSeek-V4-Flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.14,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-V4-Flash\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3-0324": {
"id": "deepseek-ai/DeepSeek-V3-0324",
"name": "DeepSeek V3 0324",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-03-24",
"last_updated": "2025-03-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 0.27,
"output": 1.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-V3-0324\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3-0324\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Flash-Vision-Exp": {
"id": "deepseek-ai/DeepSeek-V4-Flash-Vision-Exp",
"name": "DeepSeek V4 Flash Vision Exp",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-21",
"last_updated": "2026-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.44,
"output": 1.32
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-V4-Flash-Vision-Exp\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash-Vision-Exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Flash-0731": {
"id": "deepseek-ai/DeepSeek-V4-Flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.14,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-V4-Flash-0731\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.1": {
"id": "deepseek-ai/DeepSeek-V3.1",
"name": "DeepSeek-V3.1",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.27,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-V3.1\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-R1-0528": {
"id": "deepseek-ai/DeepSeek-R1-0528",
"name": "DeepSeek-R1-0528",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-05-28",
"last_updated": "2025-05-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 3,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-R1-0528\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-R1-0528\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4.1-Flash": {
"id": "deepseek-ai/DeepSeek-V4.1-Flash",
"name": "DeepSeek V4.1 Flash",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-V4.1-Flash\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4.1-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro-0813": {
"id": "deepseek-ai/DeepSeek-V4-Pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 1.32,
"output": 3.96
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-V4-Pro-0813\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-R1": {
"id": "deepseek-ai/DeepSeek-R1",
"name": "DeepSeek-R1",
"description": "Classic open reasoning model for transparent math, coding, and deliberate problem solving",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-01-20",
"last_updated": "2025-05-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 64000,
"output": 32768
},
"cost": {
"input": 0.7,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-R1\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-R1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.2": {
"id": "deepseek-ai/DeepSeek-V3.2",
"name": "DeepSeek-V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 65536
},
"cost": {
"input": 0.28,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-V3.2\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro": {
"id": "deepseek-ai/DeepSeek-V4-Pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393216
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.003625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/deepseek-ai/DeepSeek-V4-Pro\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-4b-it": {
"id": "google/gemma-3-4b-it",
"name": "Gemma 3 4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-03-12",
"last_updated": "2025-03-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/google/gemma-3-4b-it\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"google/gemma-3-4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31B-it": {
"id": "google/gemma-4-31B-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.14,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/google/gemma-4-31B-it\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"google/gemma-4-31B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-27b-it": {
"id": "google/gemma-3-27b-it",
"name": "Gemma 3 27B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-03-12",
"last_updated": "2025-03-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.08,
"output": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/google/gemma-3-27b-it\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"google/gemma-3-27b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-26B-A4B-it": {
"id": "google/gemma-4-26B-A4B-it",
"name": "Gemma 4 26B A4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.13,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/google/gemma-4-26B-A4B-it\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"google/gemma-4-26B-A4B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-12b-it": {
"id": "google/gemma-3-12b-it",
"name": "Gemma 3 12B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-03-12",
"last_updated": "2025-03-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/google/gemma-3-12b-it\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"google/gemma-3-12b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.6V-Flash": {
"id": "zai-org/GLM-4.6V-Flash",
"name": "GLM-4.6V-Flash",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-4.6V-Flash\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.6V-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.1": {
"id": "zai-org/GLM-5.1",
"name": "GLM-5.1",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-04-03",
"last_updated": "2026-04-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-5.1\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.5V": {
"id": "zai-org/GLM-4.5V",
"name": "GLM-4.5V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-08-11",
"last_updated": "2025-08-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 16384
},
"cost": {
"input": 0.6,
"output": 1.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-4.5V\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.5V\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.5": {
"id": "zai-org/GLM-4.5",
"name": "GLM-4.5",
"description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0.6,
"output": 2.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-4.5\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.3": {
"id": "zai-org/GLM-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-5.3\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.5-Air": {
"id": "zai-org/GLM-4.5-Air",
"name": "GLM-4.5-Air",
"description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
"family": "glm-air",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0.13,
"output": 0.85
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-4.5-Air\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.5-Air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2": {
"id": "zai-org/GLM-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-5.2\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.7-Flash": {
"id": "zai-org/GLM-4.7-Flash",
"name": "GLM-4.7-Flash",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-08-08",
"last_updated": "2025-08-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-4.7-Flash\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.7-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.7": {
"id": "zai-org/GLM-4.7",
"name": "GLM-4.7",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-4.7\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5": {
"id": "zai-org/GLM-5",
"name": "GLM-5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-11",
"last_updated": "2026-02-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-5\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.6": {
"id": "zai-org/GLM-4.6",
"name": "GLM-4.6",
"description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.55,
"output": 2.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-4.6\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.3-Flash": {
"id": "zai-org/GLM-5.3-Flash",
"name": "GLM-5.3-Flash",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/zai-org/GLM-5.3-Flash\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.3-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/Inkling-Small": {
"id": "thinkingmachines/Inkling-Small",
"name": "Inkling Small",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-30",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 1048576
},
"cost": {
"input": 0.5,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/thinkingmachines/Inkling-Small\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"thinkingmachines/Inkling-Small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/Inkling": {
"id": "thinkingmachines/Inkling",
"name": "Inkling",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1,
"output": 4.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/thinkingmachines/Inkling\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"thinkingmachines/Inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.8-27B": {
"id": "Qwen/Qwen3.8-27B",
"name": "Qwen3.8 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3.8-27B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.8-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-27B": {
"id": "Qwen/Qwen3.5-27B",
"name": "Qwen3.5 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3.5-27B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Coder-30B-A3B-Instruct": {
"id": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
"name": "Qwen3-Coder 30B-A3B Instruct",
"description": "Smaller Qwen coder for efficient local agents and repo-level fixes",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.07,
"output": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-Coder-30B-A3B-Instruct\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Coder-30B-A3B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-9B": {
"id": "Qwen/Qwen3.5-9B",
"name": "Qwen3.5 9B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.17,
"output": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3.5-9B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-9B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Coder-Next": {
"id": "Qwen/Qwen3-Coder-Next",
"name": "Qwen3-Coder-Next",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-03",
"last_updated": "2026-02-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.2,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-Coder-Next\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Coder-Next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-30B-A3B": {
"id": "Qwen/Qwen3-30B-A3B",
"name": "Qwen3 30B A3B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-28",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 16384
},
"cost": {
"input": 0.12,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-30B-A3B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-30B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B": {
"id": "Qwen/Qwen3-235B-A22B",
"name": "Qwen3 235B-A22B",
"description": "Large open Qwen MoE for multilingual reasoning, coding, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 16384
},
"cost": {
"input": 0.2,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-235B-A22B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-122B-A10B": {
"id": "Qwen/Qwen3.5-122B-A10B",
"name": "Qwen3.5 122B-A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3.5-122B-A10B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-122B-A10B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.8-2.4T-A95B": {
"id": "Qwen/Qwen3.8-2.4T-A95B",
"name": "Qwen3.8 2.4T A95B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 2.5,
"output": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3.8-2.4T-A95B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.8-2.4T-A95B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B-Instruct-2507": {
"id": "Qwen/Qwen3-235B-A22B-Instruct-2507",
"name": "Qwen3 235B-A22B Instruct 2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-21",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.855,
"output": 2.565
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-235B-A22B-Instruct-2507\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-235B-A22B-Instruct": {
"id": "Qwen/Qwen3-VL-235B-A22B-Instruct",
"name": "Qwen3 VL 235B A22B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-VL-235B-A22B-Instruct\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-235B-A22B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Coder-480B-A35B-Instruct": {
"id": "Qwen/Qwen3-Coder-480B-A35B-Instruct",
"name": "Qwen3-Coder-480B-A35B-Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 66536
},
"cost": {
"input": 2,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-Coder-480B-A35B-Instruct\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Coder-480B-A35B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Next-80B-A3B-Instruct": {
"id": "Qwen/Qwen3-Next-80B-A3B-Instruct",
"name": "Qwen3-Next-80B-A3B-Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-11",
"last_updated": "2025-09-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 66536
},
"cost": {
"input": 0.25,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-Next-80B-A3B-Instruct\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Next-80B-A3B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-397B-A17B": {
"id": "Qwen/Qwen3.5-397B-A17B",
"name": "Qwen3.5-397B-A17B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-01",
"last_updated": "2026-02-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3.5-397B-A17B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-397B-A17B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-35B-A3B": {
"id": "Qwen/Qwen3.5-35B-A3B",
"name": "Qwen3.5 35B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3.5-35B-A3B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-32B": {
"id": "Qwen/Qwen3-32B",
"name": "Qwen3 32B",
"description": "Dense open Qwen model for self-hosted chat, reasoning, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.29,
"output": 0.59
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-32B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-32B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B-Thinking-2507": {
"id": "Qwen/Qwen3-235B-A22B-Thinking-2507",
"name": "Qwen3-235B-A22B-Thinking-2507",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-25",
"last_updated": "2025-07-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-235B-A22B-Thinking-2507\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B-Thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-27B": {
"id": "Qwen/Qwen3.6-27B",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.47,
"output": 3.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3.6-27B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-35B-A3B": {
"id": "Qwen/Qwen3.6-35B-A3B",
"name": "Qwen3.6 35B-A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 0.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3.6-35B-A3B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-235B-A22B-Thinking": {
"id": "Qwen/Qwen3-VL-235B-A22B-Thinking",
"name": "Qwen3 VL 235B A22B Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.98,
"output": 3.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-VL-235B-A22B-Thinking\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-235B-A22B-Thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Embedding-8B": {
"id": "Qwen/Qwen3-Embedding-8B",
"name": "Qwen 3 Embedding 8B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2024-12",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 4096
},
"cost": {
"input": 0.01,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-Embedding-8B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Embedding-8B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Next-80B-A3B-Thinking": {
"id": "Qwen/Qwen3-Next-80B-A3B-Thinking",
"name": "Qwen3-Next-80B-A3B-Thinking",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-11",
"last_updated": "2025-09-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-Next-80B-A3B-Thinking\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Next-80B-A3B-Thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Embedding-4B": {
"id": "Qwen/Qwen3-Embedding-4B",
"name": "Qwen 3 Embedding 4B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2024-12",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 2048
},
"cost": {
"input": 0.01,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen3-Embedding-4B\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Embedding-4B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen2.5-Coder-32B-Instruct": {
"id": "Qwen/Qwen2.5-Coder-32B-Instruct",
"name": "Qwen2.5-Coder-32B-Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-11-12",
"last_updated": "2024-11-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.06,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/Qwen/Qwen2.5-Coder-32B-Instruct\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen2.5-Coder-32B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M2": {
"id": "MiniMaxAI/MiniMax-M2",
"name": "MiniMax-M2",
"description": "Efficient open MiniMax model built for coding agents and tool-heavy workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-10-27",
"last_updated": "2025-10-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/MiniMaxAI/MiniMax-M2\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M2.1": {
"id": "MiniMaxAI/MiniMax-M2.1",
"name": "MiniMax-M2.1",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-10",
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/MiniMaxAI/MiniMax-M2.1\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M2.5": {
"id": "MiniMaxAI/MiniMax-M2.5",
"name": "MiniMax-M2.5",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/MiniMaxAI/MiniMax-M2.5\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M3": {
"id": "MiniMaxAI/MiniMax-M3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 512000
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/MiniMaxAI/MiniMax-M3\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M2.7": {
"id": "MiniMaxAI/MiniMax-M2.7",
"name": "MiniMax-M2.7",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/MiniMaxAI/MiniMax-M2.7\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.1-8B-Instruct": {
"id": "meta-llama/Llama-3.1-8B-Instruct",
"name": "Llama-3.1-8B-Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 4096
},
"cost": {
"input": 0.06,
"output": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/meta-llama/Llama-3.1-8B-Instruct\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.1-8B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.3-70B-Instruct": {
"id": "meta-llama/Llama-3.3-70B-Instruct",
"name": "Llama-3.3-70B-Instruct",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 4096
},
"cost": {
"input": 0.59,
"output": 0.79
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/meta-llama/Llama-3.3-70B-Instruct\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.3-70B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/openai/gpt-oss-20b\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.25,
"output": 0.69
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/openai/gpt-oss-120b\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2-Thinking": {
"id": "moonshotai/Kimi-K2-Thinking",
"name": "Kimi-K2-Thinking",
"description": "Kimi reasoning model for long-horizon research, planning, and tool use",
"family": "kimi-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-11-06",
"last_updated": "2025-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 2.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/moonshotai/Kimi-K2-Thinking\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2-Thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2-Instruct-0905": {
"id": "moonshotai/Kimi-K2-Instruct-0905",
"name": "Kimi-K2-Instruct-0905",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-09-04",
"last_updated": "2025-09-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 1,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/moonshotai/Kimi-K2-Instruct-0905\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2-Instruct-0905\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2-Instruct": {
"id": "moonshotai/Kimi-K2-Instruct",
"name": "Kimi-K2-Instruct",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-07-14",
"last_updated": "2025-07-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 1,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/moonshotai/Kimi-K2-Instruct\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.5": {
"id": "moonshotai/Kimi-K2.5",
"name": "Kimi-K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-01",
"last_updated": "2026-01-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/moonshotai/Kimi-K2.5\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.7-Code": {
"id": "moonshotai/Kimi-K2.7-Code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/moonshotai/Kimi-K2.7-Code\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.7-Code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6": {
"id": "moonshotai/Kimi-K2.6",
"name": "Kimi-K2.6",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-20",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K3": {
"id": "moonshotai/Kimi-K3",
"name": "Kimi K3",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 3,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/moonshotai/Kimi-K3\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/Hy3": {
"id": "tencent/Hy3",
"name": "Hy3",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 192000,
"output": 128000
},
"cost": {
"input": 0.14,
"output": 0.58
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/tencent/Hy3\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"tencent/Hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"XiaomiMiMo/MiMo-V2.5-Pro": {
"id": "XiaomiMiMo/MiMo-V2.5-Pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/XiaomiMiMo/MiMo-V2.5-Pro\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"XiaomiMiMo/MiMo-V2.5-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"XiaomiMiMo/MiMo-V2.5": {
"id": "XiaomiMiMo/MiMo-V2.5",
"name": "MiMo-V2.5",
"description": "MiMo model for long-context reasoning, perception, and agentic tasks",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.4,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/XiaomiMiMo/MiMo-V2.5\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"XiaomiMiMo/MiMo-V2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"XiaomiMiMo/MiMo-V2-Flash": {
"id": "XiaomiMiMo/MiMo-V2-Flash",
"name": "MiMo-V2-Flash",
"description": "MiMo flash model for fast multimodal assistance and agent workflows",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-12-16",
"last_updated": "2025-12-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 4096
},
"cost": {
"input": 0.1,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"huggingface/XiaomiMiMo/MiMo-V2-Flash\", apiKey: processEnvironment[\"HF_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://router.huggingface.co/v1\")!,\n apiKey: processEnvironment[\"HF_TOKEN\"]\n)\nlet session = provider.model(\"XiaomiMiMo/MiMo-V2-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"zhipuai-coding-plan": {
"id": "zhipuai-coding-plan",
"name": "Zhipu AI Coding Plan",
"baseURL": "https://open.bigmodel.cn/api/coding/paas/v4",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ZHIPU_API_KEY"
],
"doc": "https://docs.bigmodel.cn/cn/coding-plan/overview",
"modelCount": 10,
"models": {
"glm-5.3-highspeed": {
"id": "glm-5.3-highspeed",
"name": "GLM-5.3 Highspeed",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai-coding-plan/glm-5.3-highspeed\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai-coding-plan/glm-5.3-flash\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-27",
"last_updated": "2026-03-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai-coding-plan/glm-5.1\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai-coding-plan/glm-5.3\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5v-turbo": {
"id": "glm-5v-turbo",
"name": "GLM-5V-Turbo",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-04-01",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai-coding-plan/glm-5v-turbo\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5v-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5-turbo": {
"id": "glm-5-turbo",
"name": "GLM-5-Turbo",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai-coding-plan/glm-5-turbo\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai-coding-plan/glm-5.2\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.6v": {
"id": "glm-4.6v",
"name": "GLM-4.6V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai-coding-plan/glm-4.6v\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.6v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.7": {
"id": "glm-4.7",
"name": "GLM-4.7",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai-coding-plan/glm-4.7\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-highspeed": {
"id": "glm-5.2-highspeed",
"name": "GLM-5.2 Highspeed",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai-coding-plan/glm-5.2-highspeed\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/coding/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"daoxe": {
"id": "daoxe",
"name": "DaoXE",
"baseURL": "https://daoxe.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"DAOXE_API_KEY"
],
"doc": "https://daoxe.com/pricing",
"modelCount": 9,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"daoxe/claude-sonnet-4-6\", apiKey: processEnvironment[\"DAOXE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://daoxe.com/v1\")!,\n apiKey: processEnvironment[\"DAOXE_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.3": {
"id": "grok-4.3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"daoxe/grok-4.3\", apiKey: processEnvironment[\"DAOXE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://daoxe.com/v1\")!,\n apiKey: processEnvironment[\"DAOXE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro-preview": {
"id": "gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"daoxe/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"DAOXE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://daoxe.com/v1\")!,\n apiKey: processEnvironment[\"DAOXE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.5": {
"id": "grok-4.5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"daoxe/grok-4.5\", apiKey: processEnvironment[\"DAOXE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://daoxe.com/v1\")!,\n apiKey: processEnvironment[\"DAOXE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"daoxe/gpt-5.4\", apiKey: processEnvironment[\"DAOXE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://daoxe.com/v1\")!,\n apiKey: processEnvironment[\"DAOXE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5-20251001": {
"id": "claude-haiku-4-5-20251001",
"name": "Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"daoxe/claude-haiku-4-5-20251001\", apiKey: processEnvironment[\"DAOXE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://daoxe.com/v1\")!,\n apiKey: processEnvironment[\"DAOXE_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5-20251001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"daoxe/kimi-k2.5\", apiKey: processEnvironment[\"DAOXE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://daoxe.com/v1\")!,\n apiKey: processEnvironment[\"DAOXE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"daoxe/claude-opus-4-8\", apiKey: processEnvironment[\"DAOXE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://daoxe.com/v1\")!,\n apiKey: processEnvironment[\"DAOXE_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"daoxe/gpt-5.5\", apiKey: processEnvironment[\"DAOXE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://daoxe.com/v1\")!,\n apiKey: processEnvironment[\"DAOXE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"crossmodel": {
"id": "crossmodel",
"name": "CrossModel",
"baseURL": "https://api.crossmodel.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"CROSSMODEL_API_KEY"
],
"doc": "https://www.crossmodel.ai/docs",
"modelCount": 59,
"models": {
"qwen/qwen3.7-max": {
"id": "qwen/qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 1.88,
"output": 5.63,
"cache_read": 0.375,
"cache_write": 2.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/qwen/qwen3.7-max\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-plus": {
"id": "qwen/qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.32,
"output": 1.88,
"cache_read": 0.032,
"cache_write": 0.4,
"tiers": [
{
"input": 1.25,
"output": 7.5,
"cache_read": 0.124,
"cache_write": 1.57,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 1.25,
"output": 7.5,
"cache_read": 0.124,
"cache_write": 1.57
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/qwen/qwen3.6-plus\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.7-flash": {
"id": "qwen/qwen3.7-flash",
"name": "Qwen3.7 Flash",
"description": "Lightweight multimodal Qwen model for high-throughput text, image, and video tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 991000,
"output": 65536
},
"cost": {
"input": 0.04,
"output": 0.13,
"cache_read": 0.01,
"cache_write": 0.04,
"tiers": [
{
"input": 0.1,
"output": 0.37,
"cache_read": 0.02,
"cache_write": 0.12,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.19,
"output": 0.74,
"cache_read": 0.04,
"cache_write": 0.24,
"tier": {
"type": "context",
"size": 256000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/qwen/qwen3.7-flash\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-flash": {
"id": "qwen/qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.19,
"output": 1.13,
"cache_read": 0.019,
"cache_write": 0.24,
"tiers": [
{
"input": 0.75,
"output": 4.5,
"cache_read": 0.075,
"cache_write": 0.94,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075,
"cache_write": 0.94
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/qwen/qwen3.6-flash\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-flash": {
"id": "qwen/qwen3.8-flash",
"name": "Qwen3.8 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.13,
"output": 0.43,
"cache_read": 0.016,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/qwen/qwen3.8-flash\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-max": {
"id": "qwen/qwen3.8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.88,
"output": 5.63,
"cache_read": 0.23,
"cache_write": 2.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/qwen/qwen3.8-max\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.7-plus": {
"id": "qwen/qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.32,
"output": 1.25,
"cache_read": 0.032,
"cache_write": 0.4,
"tiers": [
{
"input": 0.96,
"output": 3.75,
"cache_read": 0.096,
"cache_write": 1.2,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 0.96,
"output": 3.75,
"cache_read": 0.096,
"cache_write": 1.2
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/qwen/qwen3.7-plus\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5": {
"id": "xiaomi/mimo-v2.5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.16,
"output": 0.32,
"cache_read": 0.004,
"cache_write": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/xiaomi/mimo-v2.5\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5-pro": {
"id": "xiaomi/mimo-v2.5-pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.47,
"output": 0.94,
"cache_read": 0.005,
"cache_write": 0.47
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/xiaomi/mimo-v2.5-pro\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7": {
"id": "minimax/minimax-m2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.33,
"output": 1.32,
"cache_read": 0.066,
"cache_write": 0.42
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/minimax/minimax-m2.7\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m3": {
"id": "minimax/minimax-m3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1024000,
"output": 512000
},
"cost": {
"input": 0.33,
"output": 1.32,
"cache_read": 0.066,
"cache_write": 0.33,
"tiers": [
{
"input": 0.66,
"output": 2.63,
"cache_read": 0.132,
"cache_write": 0.66,
"tier": {
"type": "context",
"size": 512000
}
}
],
"context_over_200k": {
"input": 0.66,
"output": 2.63,
"cache_read": 0.132,
"cache_write": 0.66
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/minimax/minimax-m3\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4-6": {
"id": "anthropic/claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/anthropic/claude-sonnet-4-6\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-5": {
"id": "anthropic/claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/anthropic/claude-opus-5\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5-1": {
"id": "anthropic/claude-fable-5-1",
"name": "Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/anthropic/claude-fable-5-1\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-7": {
"id": "anthropic/claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/anthropic/claude-opus-4-7\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5": {
"id": "anthropic/claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/anthropic/claude-fable-5\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-4-5": {
"id": "anthropic/claude-haiku-4-5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/anthropic/claude-haiku-4-5\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-8": {
"id": "anthropic/claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/anthropic/claude-opus-4-8\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-5": {
"id": "anthropic/claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/anthropic/claude-sonnet-5\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot/kimi-k2.6": {
"id": "moonshot/kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 1,
"output": 4.16,
"cache_read": 0.18,
"cache_write": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/moonshot/kimi-k2.6\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"moonshot/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot/kimi-k2.7-code": {
"id": "moonshot/kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 1,
"output": 4.16,
"cache_read": 0.18,
"cache_write": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/moonshot/kimi-k2.7-code\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"moonshot/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot/kimi-k3": {
"id": "moonshot/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/moonshot/kimi-k3\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"moonshot/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-vision-exp": {
"id": "deepseek/deepseek-v4-flash-vision-exp",
"name": "DeepSeek V4 Flash Vision Exp",
"description": "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-21",
"last_updated": "2026-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.27,
"output": 1.08,
"cache_read": 0.0054,
"cache_write": 0.27
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/deepseek/deepseek-v4-flash-vision-exp\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-vision-exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash": {
"id": "deepseek/deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.27,
"output": 1.08,
"cache_read": 0.0054,
"cache_write": 0.27
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/deepseek/deepseek-v4-flash\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4.1-flash": {
"id": "deepseek/deepseek-v4.1-flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.27,
"output": 1.08,
"cache_read": 0.0054,
"cache_write": 0.27
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/deepseek/deepseek-v4.1-flash\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4.1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro": {
"id": "deepseek/deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 1.215,
"output": 3.645,
"cache_read": 0.0405,
"cache_write": 1.215
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/deepseek/deepseek-v4-pro\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-3.1-pro-preview": {
"id": "gemini/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/gemini/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-2.5-flash-lite": {
"id": "gemini/gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash-Lite",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/gemini/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-3.6-flash": {
"id": "gemini/gemini-3.6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/gemini/gemini-3.6-flash\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-3.5-flash": {
"id": "gemini/gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"cache_write": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/gemini/gemini-3.5-flash\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-3.5-flash-lite": {
"id": "gemini/gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"cache_write": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/gemini/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-3-flash-preview": {
"id": "gemini/gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/gemini/gemini-3-flash-preview\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-3.8-flash": {
"id": "gemini/gemini-3.8-flash",
"name": "Gemini 3.8 Flash",
"description": "Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/gemini/gemini-3.8-flash\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-3.7-flash": {
"id": "gemini/gemini-3.7-flash",
"name": "Gemini 3.7 Flash",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/gemini/gemini-3.7-flash\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-2.5-pro": {
"id": "gemini/gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"cache_write": 1.25,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"cache_write": 2.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/gemini/gemini-2.5-pro\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-2.5-flash": {
"id": "gemini/gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"cache_write": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/gemini/gemini-2.5-flash\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.3": {
"id": "x-ai/grok-4.3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"cache_write": 1.25,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"cache_write": 2.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/x-ai/grok-4.3\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.5": {
"id": "x-ai/grok-4.5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.3,
"cache_write": 2,
"tiers": [
{
"input": 4,
"output": 12,
"cache_read": 0.6,
"cache_write": 4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 12,
"cache_read": 0.6,
"cache_write": 4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/x-ai/grok-4.5\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-build-0.1": {
"id": "x-ai/grok-build-0.1",
"name": "Grok Build 0.1",
"description": "Fast Grok coding model tuned for agentic engineering and iterative edits",
"family": "grok-build",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 1,
"output": 2,
"cache_read": 0.2,
"cache_write": 1,
"tiers": [
{
"input": 2,
"output": 4,
"cache_read": 0.4,
"cache_write": 2,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2,
"output": 4,
"cache_read": 0.4,
"cache_write": 2
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/x-ai/grok-build-0.1\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-build-0.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.6": {
"id": "x-ai/grok-4.6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5,
"cache_write": 2,
"tiers": [
{
"input": 4,
"output": 12,
"cache_read": 1,
"cache_write": 4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 12,
"cache_read": 1,
"cache_write": 4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/x-ai/grok-4.6\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-sol": {
"id": "openai/gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 4,
"output": 20,
"cache_read": 0.4,
"cache_write": 5,
"tiers": [
{
"input": 8,
"output": 30,
"cache_read": 0.8,
"cache_write": 10,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 8,
"output": 30,
"cache_read": 0.8,
"cache_write": 10
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/openai/gpt-5.6-sol\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-6-astra": {
"id": "openai/gpt-6-astra",
"name": "GPT-6 Astra",
"description": "GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.",
"family": "gpt-astra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-04-30",
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5,
"tiers": [
{
"input": 20,
"output": 75,
"cache_read": 2,
"cache_write": 25,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 20,
"output": 75,
"cache_read": 2,
"cache_write": 25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/openai/gpt-6-astra\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-6-astra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4": {
"id": "openai/gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"cache_write": 2.5,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"cache_write": 5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"cache_write": 5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/openai/gpt-5.4\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-luna": {
"id": "openai/gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25,
"tiers": [
{
"input": 0.4,
"output": 1.8,
"cache_read": 0.04,
"cache_write": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.04,
"cache_write": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/openai/gpt-5.6-luna\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini": {
"id": "openai/gpt-4o-mini",
"name": "GPT-4o mini",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/openai/gpt-4o-mini\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-nano": {
"id": "openai/gpt-5.4-nano",
"name": "GPT-5.4 nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/openai/gpt-5.4-nano\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5-pro": {
"id": "openai/gpt-5.5-pro",
"name": "GPT-5.5 Pro",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180,
"tiers": [
{
"input": 60,
"output": 270,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 60,
"output": 270
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/openai/gpt-5.5-pro\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-mini": {
"id": "openai/gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075,
"cache_write": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/openai/gpt-5.4-mini\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-terra": {
"id": "openai/gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/openai/gpt-5.6-terra\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"cache_write": 10,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1,
"cache_write": 10
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/openai/gpt-5.5\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/hy3": {
"id": "tencent/hy3",
"name": "Hy3",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 192000,
"output": 131072
},
"cost": {
"input": 0.16,
"output": 0.64,
"cache_read": 0.04,
"cache_write": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/tencent/hy3\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"tencent/hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/hy4-preview": {
"id": "tencent/hy4-preview",
"name": "Hy4 preview",
"description": "A next-generation productivity model with significantly enhanced Agent and complex task execution capabilities.",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-08-28",
"last_updated": "2026-08-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.96,
"output": 2.88,
"cache_read": 0.048,
"cache_write": 0.96
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/tencent/hy4-preview\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"tencent/hy4-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7": {
"id": "z-ai/glm-4.7",
"name": "GLM-4.7",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.47,
"output": 2.16,
"cache_read": 0.1,
"cache_write": 0.47,
"tiers": [
{
"input": 0.62,
"output": 2.47,
"cache_read": 0.13,
"cache_write": 0.62,
"tier": {
"type": "context",
"size": 32000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/z-ai/glm-4.7\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.2": {
"id": "z-ai/glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.2,
"output": 4.4,
"cache_read": 0.3,
"cache_write": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/z-ai/glm-5.2\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.3-flash": {
"id": "z-ai/glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.03,
"cache_write": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/z-ai/glm-5.3-flash\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5": {
"id": "z-ai/glm-5",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.16,
"cache_write": 0.6,
"tiers": [
{
"input": 0.8,
"output": 3.4,
"cache_read": 0.2,
"cache_write": 0.8,
"tier": {
"type": "context",
"size": 32000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/z-ai/glm-5\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.1": {
"id": "z-ai/glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 1,
"output": 3.8,
"cache_read": 0.2,
"cache_write": 1,
"tiers": [
{
"input": 1.2,
"output": 4.4,
"cache_read": 0.3,
"cache_write": 1.2,
"tier": {
"type": "context",
"size": 32000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/z-ai/glm-5.1\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5-turbo": {
"id": "z-ai/glm-5-turbo",
"name": "GLM-5-Turbo",
"description": "Faster GLM-5 lane for coding agents that need lower latency",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.9,
"output": 3.7,
"cache_read": 0.18,
"cache_write": 0.9,
"tiers": [
{
"input": 1.1,
"output": 4.3,
"cache_read": 0.27,
"cache_write": 1.1,
"tier": {
"type": "context",
"size": 32000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/z-ai/glm-5-turbo\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.3": {
"id": "z-ai/glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.2,
"output": 4.4,
"cache_read": 0.3,
"cache_write": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crossmodel/z-ai/glm-5.3\", apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.crossmodel.ai/v1\")!,\n apiKey: processEnvironment[\"CROSSMODEL_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"minimax": {
"id": "minimax",
"name": "MiniMax (minimax.io)",
"baseURL": "https://api.minimax.io/anthropic/v1",
"npm": "@ai-sdk/anthropic",
"swiftDriver": "anthropicMessages",
"env": [
"MINIMAX_API_KEY"
],
"doc": "https://platform.minimax.io/docs/guides/quickstart",
"modelCount": 7,
"models": {
"MiniMax-M2": {
"id": "MiniMax-M2",
"name": "MiniMax-M2",
"description": "Efficient open MiniMax model built for coding agents and tool-heavy workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-10-27",
"last_updated": "2025-10-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax/MiniMax-M2\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.1": {
"id": "MiniMax-M2.1",
"name": "MiniMax-M2.1",
"description": "Earlier MiniMax agent model for practical coding and productivity tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03,
"cache_write": 0.375
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax/MiniMax-M2.1\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.5": {
"id": "MiniMax-M2.5",
"name": "MiniMax-M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03,
"cache_write": 0.375
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax/MiniMax-M2.5\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.5-highspeed": {
"id": "MiniMax-M2.5-highspeed",
"name": "MiniMax-M2.5-highspeed",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-13",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax/MiniMax-M2.5-highspeed\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.5-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M3": {
"id": "MiniMax-M3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-25",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 512000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06,
"tiers": [
{
"input": 0.6,
"output": 2.4,
"cache_read": 0.12,
"tier": {
"type": "context",
"size": 512000
}
}
],
"context_over_200k": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.12
}
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax/MiniMax-M3\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.7-highspeed": {
"id": "MiniMax-M2.7-highspeed",
"name": "MiniMax-M2.7-highspeed",
"description": "Low-latency M2.7 variant for interactive coding plans and agent loops",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax/MiniMax-M2.7-highspeed\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.7": {
"id": "MiniMax-M2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax/MiniMax-M2.7\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"salad-cloud": {
"id": "salad-cloud",
"name": "SaladCloud AI Gateway",
"baseURL": "",
"npm": "@saladtechnologies-oss/ai-sdk-provider",
"swiftDriver": "openaiChat",
"env": [
"SALAD_CLOUD_API_KEY"
],
"doc": "https://docs.salad.com/ai-gateway/explanation/overview",
"modelCount": 1,
"models": {
"qwen3.6-35b-a3b": {
"id": "qwen3.6-35b-a3b",
"name": "Qwen3.6 35B-A3B",
"description": "Qwen MoE for agentic tasks, complex reasoning, code generation, and instruction following",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 262144,
"output": 262144
},
"cost": {
"input": 0.09,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"salad-cloud/qwen3.6-35b-a3b\", apiKey: processEnvironment[\"SALAD_CLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"SALAD_CLOUD_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"aki-io": {
"id": "aki-io",
"name": "AKI.IO",
"baseURL": "https://aki.io/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"AKI_IO_API_KEY"
],
"doc": "https://aki.io/docs/",
"modelCount": 7,
"models": {
"gemma4-26b": {
"id": "gemma4-26b",
"name": "Gemma 4 26B A4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aki-io/gemma4-26b\", apiKey: processEnvironment[\"AKI_IO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://aki.io/v1\")!,\n apiKey: processEnvironment[\"AKI_IO_API_KEY\"]\n)\nlet session = provider.model(\"gemma4-26b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-27b": {
"id": "qwen3.8-27b",
"name": "Qwen3.8 27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 2.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aki-io/qwen3.8-27b\", apiKey: processEnvironment[\"AKI_IO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://aki.io/v1\")!,\n apiKey: processEnvironment[\"AKI_IO_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731-284b": {
"id": "deepseek-v4-flash-0731-284b",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 81920
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aki-io/deepseek-v4-flash-0731-284b\", apiKey: processEnvironment[\"AKI_IO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://aki.io/v1\")!,\n apiKey: processEnvironment[\"AKI_IO_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731-284b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral4-119b": {
"id": "mistral4-119b",
"name": "Mistral Small 4",
"description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents",
"family": "mistral-small",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 81920
},
"cost": {
"input": 0.2,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aki-io/mistral4-119b\", apiKey: processEnvironment[\"AKI_IO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://aki.io/v1\")!,\n apiKey: processEnvironment[\"AKI_IO_API_KEY\"]\n)\nlet session = provider.model(\"mistral4-119b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code-1100b": {
"id": "kimi-k2.7-code-1100b",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 81920
},
"cost": {
"input": 0.86,
"output": 3,
"cache_read": 0.18
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aki-io/kimi-k2.7-code-1100b\", apiKey: processEnvironment[\"AKI_IO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://aki.io/v1\")!,\n apiKey: processEnvironment[\"AKI_IO_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code-1100b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aki-io/gpt-oss-120b\", apiKey: processEnvironment[\"AKI_IO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://aki.io/v1\")!,\n apiKey: processEnvironment[\"AKI_IO_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-35b": {
"id": "qwen3.6-35b",
"name": "Qwen3.6 35B-A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"aki-io/qwen3.6-35b\", apiKey: processEnvironment[\"AKI_IO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://aki.io/v1\")!,\n apiKey: processEnvironment[\"AKI_IO_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-35b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"trustedrouter": {
"id": "trustedrouter",
"name": "TrustedRouter",
"baseURL": "https://api.trustedrouter.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"TRUSTEDROUTER_API_KEY"
],
"doc": "https://trustedrouter.com/docs",
"modelCount": 7,
"models": {
"trustedrouter/zdr": {
"id": "trustedrouter/zdr",
"name": "Zero Data Retention",
"description": "TrustedRouter privacy routing alias that prefers zero data retention model endpoints.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-27",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"trustedrouter/trustedrouter/zdr\", apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.trustedrouter.com/v1\")!,\n apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"]\n)\nlet session = provider.model(\"trustedrouter/zdr\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"trustedrouter/synth": {
"id": "trustedrouter/synth",
"name": "Synth",
"description": "TrustedRouter synthesis orchestration alias that combines multiple model responses into one answer.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-20",
"last_updated": "2026-06-27",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"trustedrouter/trustedrouter/synth\", apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.trustedrouter.com/v1\")!,\n apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"]\n)\nlet session = provider.model(\"trustedrouter/synth\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"trustedrouter/e2e": {
"id": "trustedrouter/e2e",
"name": "End-to-End Encrypted",
"description": "TrustedRouter privacy routing alias for end-to-end encrypted provider routes where available.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-27",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"trustedrouter/trustedrouter/e2e\", apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.trustedrouter.com/v1\")!,\n apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"]\n)\nlet session = provider.model(\"trustedrouter/e2e\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"trustedrouter/synth-code": {
"id": "trustedrouter/synth-code",
"name": "Synth Code",
"description": "TrustedRouter code synthesis orchestration alias that combines multiple model responses into one answer.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-20",
"last_updated": "2026-06-27",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"trustedrouter/trustedrouter/synth-code\", apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.trustedrouter.com/v1\")!,\n apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"]\n)\nlet session = provider.model(\"trustedrouter/synth-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"trustedrouter/fast": {
"id": "trustedrouter/fast",
"name": "Fast",
"description": "TrustedRouter speed routing alias that prefers low-latency healthy model endpoints.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-27",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"trustedrouter/trustedrouter/fast\", apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.trustedrouter.com/v1\")!,\n apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"]\n)\nlet session = provider.model(\"trustedrouter/fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"trustedrouter/cheap": {
"id": "trustedrouter/cheap",
"name": "Cheap",
"description": "TrustedRouter low-cost routing alias that prefers inexpensive healthy model endpoints.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-01",
"last_updated": "2026-06-27",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"trustedrouter/trustedrouter/cheap\", apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.trustedrouter.com/v1\")!,\n apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"]\n)\nlet session = provider.model(\"trustedrouter/cheap\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"trustedrouter/auto": {
"id": "trustedrouter/auto",
"name": "Auto",
"description": "TrustedRouter automatic routing alias that chooses a healthy supported model endpoint for the request.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-01",
"last_updated": "2026-06-27",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"trustedrouter/trustedrouter/auto\", apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.trustedrouter.com/v1\")!,\n apiKey: processEnvironment[\"TRUSTEDROUTER_API_KEY\"]\n)\nlet session = provider.model(\"trustedrouter/auto\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"alibaba": {
"id": "alibaba",
"name": "Alibaba",
"baseURL": "https://dashscope-intl.aliyuncs.com/compatible-mode/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"DASHSCOPE_API_KEY"
],
"doc": "https://www.alibabacloud.com/help/en/model-studio/models",
"modelCount": 55,
"models": {
"qwen3.7-max": {
"id": "qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.5,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.7-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-72b-instruct": {
"id": "qwen2-5-72b-instruct",
"name": "Qwen2.5 72B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 1.4,
"output": 5.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen2-5-72b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.2,
"output": 0.4,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-plus": {
"id": "qwen3-coder-plus",
"name": "Qwen3 Coder Plus",
"description": "Hosted Qwen coder for software agents, repo edits, and long-context code",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-coder-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-next-80b-a3b-thinking": {
"id": "qwen3-next-80b-a3b-thinking",
"name": "Qwen3-Next 80B-A3B (Thinking)",
"description": "Efficient Qwen thinking model for local reasoning, math, and coding agents",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-next-80b-a3b-thinking\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-next-80b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-omni-7b": {
"id": "qwen2-5-omni-7b",
"name": "Qwen2.5-Omni 7B",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-12",
"last_updated": "2024-12",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text",
"audio"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 2048
},
"cost": {
"input": 0.1,
"output": 0.4,
"input_audio": 6.76
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen2-5-omni-7b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-omni-7b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-mt-turbo": {
"id": "qwen-mt-turbo",
"name": "Qwen-MT Turbo",
"description": "Translation model for multilingual conversion, localization, and cross-language workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-01",
"last_updated": "2025-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"output": 8192
},
"cost": {
"input": 0.16,
"output": 0.49
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-mt-turbo\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-mt-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-vl-max": {
"id": "qwen-vl-max",
"name": "Qwen-VL Max",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-04-08",
"last_updated": "2025-08-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.8,
"output": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-vl-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-vl-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-next-80b-a3b-instruct": {
"id": "qwen3-next-80b-a3b-instruct",
"name": "Qwen3-Next 80B-A3B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-flash": {
"id": "qwen3-coder-flash",
"name": "Qwen3 Coder Flash",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-coder-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-14b": {
"id": "qwen3-14b",
"name": "Qwen3 14B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.35,
"output": 1.4,
"reasoning": 4.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-14b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-14b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-max": {
"id": "qwen-max",
"name": "Qwen Max",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-04-03",
"last_updated": "2025-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 1.6,
"output": 6.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-plus": {
"id": "qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"cache_write": 0.625,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.6-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-vl-plus": {
"id": "qwen-vl-plus",
"name": "Qwen-VL Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01-25",
"last_updated": "2025-08-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.21,
"output": 0.63
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-vl-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-vl-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-omni-turbo-realtime": {
"id": "qwen-omni-turbo-realtime",
"name": "Qwen-Omni Turbo Realtime",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-05-08",
"last_updated": "2025-05-08",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 2048
},
"cost": {
"input": 0.27,
"output": 1.07,
"input_audio": 4.44,
"output_audio": 8.89
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-omni-turbo-realtime\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-omni-turbo-realtime\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-27b": {
"id": "qwen3.5-27b",
"name": "Qwen3.5 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.5-27b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-35b-a3b": {
"id": "qwen3.5-35b-a3b",
"name": "Qwen3.5 35B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.5-35b-a3b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-flash": {
"id": "qwen-flash",
"name": "Qwen Flash",
"description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.28,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/glm-5.2\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-turbo": {
"id": "qwen-turbo",
"name": "Qwen Turbo",
"description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-11-01",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 16384
},
"cost": {
"input": 0.05,
"output": 0.2,
"reasoning": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-turbo\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-livetranslate-flash-realtime": {
"id": "qwen3-livetranslate-flash-realtime",
"name": "Qwen3-LiveTranslate Flash Realtime",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-09-22",
"last_updated": "2025-09-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 53248,
"output": 4096
},
"cost": {
"input": 10,
"output": 10,
"input_audio": 10,
"output_audio": 38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-livetranslate-flash-realtime\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-livetranslate-flash-realtime\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-30b-a3b": {
"id": "qwen3-vl-30b-a3b",
"name": "Qwen3-VL 30B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.8,
"reasoning": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-vl-30b-a3b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-vl-ocr": {
"id": "qwen-vl-ocr",
"name": "Qwen-VL OCR",
"description": "OCR model for extracting structured text from documents and screenshots",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-10-28",
"last_updated": "2025-04-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 34096,
"output": 4096
},
"cost": {
"input": 0.72,
"output": 0.72
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-vl-ocr\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-vl-ocr\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-32b": {
"id": "qwen3-32b",
"name": "Qwen3 32B",
"description": "Dense open Qwen model for self-hosted chat, reasoning, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.7,
"output": 2.8,
"reasoning": 8.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-32b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwq-plus": {
"id": "qwq-plus",
"name": "QwQ Plus",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-03-05",
"last_updated": "2025-03-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.8,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwq-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwq-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-plus": {
"id": "qwen3-vl-plus",
"name": "Qwen3-VL Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 1.6,
"reasoning": 4.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-vl-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-7b-instruct": {
"id": "qwen2-5-7b-instruct",
"name": "Qwen2.5 7B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.175,
"output": 0.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen2-5-7b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-30b-a3b-instruct": {
"id": "qwen3-coder-30b-a3b-instruct",
"name": "Qwen3-Coder 30B-A3B Instruct",
"description": "Smaller Qwen coder for efficient local agents and repo-level fixes",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.45,
"output": 2.25,
"tiers": [
{
"input": 0.75,
"output": 3.75,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 1.2,
"output": 6,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-397b-a17b": {
"id": "qwen3.5-397b-a17b",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-27b": {
"id": "qwen3.6-27b",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.6-27b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-asr-flash": {
"id": "qwen3-asr-flash",
"name": "Qwen3-ASR Flash",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2024-04",
"release_date": "2025-09-08",
"last_updated": "2025-09-08",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 53248,
"output": 4096
},
"cost": {
"input": 0.035,
"output": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-asr-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-asr-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-omni-flash": {
"id": "qwen3-omni-flash",
"name": "Qwen3-Omni Flash",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-09-15",
"last_updated": "2025-09-15",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 16384
},
"cost": {
"input": 0.43,
"output": 1.66,
"input_audio": 3.81,
"output_audio": 15.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-omni-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-omni-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-vl-72b-instruct": {
"id": "qwen2-5-vl-72b-instruct",
"name": "Qwen2.5-VL 72B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 2.8,
"output": 8.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen2-5-vl-72b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-vl-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-35b-a3b": {
"id": "qwen3.6-35b-a3b",
"name": "Qwen3.6 35B-A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.248,
"output": 1.485
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.6-35b-a3b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-max": {
"id": "qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 1.2,
"output": 6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-plus": {
"id": "qwen-plus",
"name": "Qwen Plus",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01-25",
"last_updated": "2025-09-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.2,
"reasoning": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-122b-a10b": {
"id": "qwen3.5-122b-a10b",
"name": "Qwen3.5 122B-A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.5-122b-a10b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-122b-a10b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-omni-flash-realtime": {
"id": "qwen3-omni-flash-realtime",
"name": "Qwen3-Omni Flash Realtime",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-09-15",
"last_updated": "2025-09-15",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 16384
},
"cost": {
"input": 0.52,
"output": 1.99,
"input_audio": 4.57,
"output_audio": 18.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-omni-flash-realtime\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-omni-flash-realtime\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-vl-7b-instruct": {
"id": "qwen2-5-vl-7b-instruct",
"name": "Qwen2.5-VL 7B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.35,
"output": 1.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen2-5-vl-7b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-vl-7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-flash": {
"id": "qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.1875,
"output": 1.125,
"cache_write": 0.234375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.6-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-flash": {
"id": "qwen3.8-flash",
"name": "Qwen3.8 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens",
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.47,
"cache_read": 0.016,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.8-flash\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-14b-instruct": {
"id": "qwen2-5-14b-instruct",
"name": "Qwen2.5 14B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.35,
"output": 1.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen2-5-14b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-14b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-max-preview": {
"id": "qwen3.6-max-preview",
"name": "Qwen3.6 Max Preview",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-20",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 1.3,
"output": 7.8,
"cache_read": 0.13,
"cache_write": 1.625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.6-max-preview\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-plus-character-ja": {
"id": "qwen-plus-character-ja",
"name": "Qwen Plus Character (Japanese)",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01",
"last_updated": "2024-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 512
},
"cost": {
"input": 0.5,
"output": 1.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-plus-character-ja\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-plus-character-ja\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max": {
"id": "qwen3.8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens",
"min": 0,
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.8-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-8b": {
"id": "qwen3-8b",
"name": "Qwen3 8B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.18,
"output": 0.7,
"reasoning": 2.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-8b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b": {
"id": "qwen3-235b-a22b",
"name": "Qwen3 235B-A22B",
"description": "Large open Qwen MoE for multilingual reasoning, coding, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.7,
"output": 2.8,
"reasoning": 8.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-235b-a22b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-plus": {
"id": "qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"cache_write": 0.625,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.7-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-omni-turbo": {
"id": "qwen-omni-turbo",
"name": "Qwen-Omni Turbo",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-01-19",
"last_updated": "2025-03-26",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 2048
},
"cost": {
"input": 0.07,
"output": 0.27,
"input_audio": 4.44,
"output_audio": 8.89
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-omni-turbo\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-omni-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-480b-a35b-instruct": {
"id": "qwen3-coder-480b-a35b-instruct",
"name": "Qwen3-Coder 480B-A35B Instruct",
"description": "Open Qwen coding heavyweight for repository reasoning and agentic engineering",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 7.5,
"tiers": [
{
"input": 2.7,
"output": 13.5,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 4.5,
"output": 22.5,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-coder-480b-a35b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-480b-a35b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2-5-32b-instruct": {
"id": "qwen2-5-32b-instruct",
"name": "Qwen2.5 32B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-09",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.7,
"output": 2.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen2-5-32b-instruct\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen2-5-32b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-235b-a22b": {
"id": "qwen3-vl-235b-a22b",
"name": "Qwen3-VL 235B-A22B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.7,
"output": 2.8,
"reasoning": 8.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3-vl-235b-a22b\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-235b-a22b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-plus": {
"id": "qwen3.5-plus",
"name": "Qwen3.5 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 2.4,
"reasoning": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen3.5-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-mt-plus": {
"id": "qwen-mt-plus",
"name": "Qwen-MT Plus",
"description": "Translation model for multilingual conversion, localization, and cross-language workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-01",
"last_updated": "2025-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"output": 8192
},
"cost": {
"input": 2.46,
"output": 7.37
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qwen-mt-plus\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-mt-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qvq-max": {
"id": "qvq-max",
"name": "QVQ Max",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qvq",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-03-25",
"last_updated": "2025-03-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 1.2,
"output": 4.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"alibaba/qvq-max\", apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://dashscope-intl.aliyuncs.com/compatible-mode/v1\")!,\n apiKey: processEnvironment[\"DASHSCOPE_API_KEY\"]\n)\nlet session = provider.model(\"qvq-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"nvidia": {
"id": "nvidia",
"name": "Nvidia",
"baseURL": "https://integrate.api.nvidia.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"NVIDIA_API_KEY"
],
"doc": "https://docs.api.nvidia.com/nim/",
"modelCount": 103,
"models": {
"qwen/qwen3-next-80b-a3b-instruct": {
"id": "qwen/qwen3-next-80b-a3b-instruct",
"name": "Qwen3-Next-80B-A3B-Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2024-12-01",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/qwen/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-397b-a17b": {
"id": "qwen/qwen3.5-397b-a17b",
"name": "Qwen3.5-397B-A17B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-01",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/qwen/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-122b-a10b": {
"id": "qwen/qwen3.5-122b-a10b",
"name": "Qwen3.5 122B-A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/qwen/qwen3.5-122b-a10b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-122b-a10b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-image": {
"id": "qwen/qwen-image",
"name": "Qwen Image",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/qwen/qwen-image\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-480b-a35b-instruct": {
"id": "qwen/qwen3-coder-480b-a35b-instruct",
"name": "Qwen3 Coder 480B A35B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 66536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/qwen/qwen3-coder-480b-a35b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-480b-a35b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-image-edit": {
"id": "qwen/qwen-image-edit",
"name": "Qwen Image Edit",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-19",
"last_updated": "2025-08-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/qwen/qwen-image-edit\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-image-edit\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen2.5-coder-32b-instruct": {
"id": "qwen/qwen2.5-coder-32b-instruct",
"name": "Qwen2.5 Coder 32b Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-11-06",
"last_updated": "2024-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/qwen/qwen2.5-coder-32b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen2.5-coder-32b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun-ai/step-3.7-flash": {
"id": "stepfun-ai/step-3.7-flash",
"name": "Step 3.7 Flash",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/stepfun-ai/step-3.7-flash\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"stepfun-ai/step-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun-ai/step-3.5-flash": {
"id": "stepfun-ai/step-3.5-flash",
"name": "Step 3.5 Flash",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-02",
"last_updated": "2026-02-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/stepfun-ai/step-3.5-flash\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"stepfun-ai/step-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/deepseek-v4-pro-0813": {
"id": "deepseek-ai/deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/deepseek-ai/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/deepseek-v4-flash-0731": {
"id": "deepseek-ai/deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/deepseek-ai/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/deepseek-v4-flash": {
"id": "deepseek-ai/deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393216
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/deepseek-ai/deepseek-v4-flash\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/deepseek-v4-pro": {
"id": "deepseek-ai/deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393216
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.003625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/deepseek-ai/deepseek-v4-pro\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-xs-2.1": {
"id": "poolside/laguna-xs-2.1",
"name": "Laguna XS 2.1",
"description": "Agentic coding model from Poolside in the XS size class for local deployment",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-02",
"last_updated": "2026-07-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/poolside/laguna-xs-2.1\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-xs-2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-large-3-675b-instruct-2512": {
"id": "mistralai/mistral-large-3-675b-instruct-2512",
"name": "Mistral Large 3 675B Instruct 2512",
"description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
"family": "mistral-large",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/mistralai/mistral-large-3-675b-instruct-2512\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-large-3-675b-instruct-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-nemotron": {
"id": "mistralai/mistral-nemotron",
"name": "mistral-nemotron",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-06-11",
"last_updated": "2025-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/mistralai/mistral-nemotron\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-nemotron\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mixtral-8x22b-instruct": {
"id": "mistralai/mixtral-8x22b-instruct",
"name": "Mistral: Mixtral 8x22B Instruct",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-04-17",
"last_updated": "2024-04-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 13108
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/mistralai/mixtral-8x22b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mixtral-8x22b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/ministral-14b-instruct-2512": {
"id": "mistralai/ministral-14b-instruct-2512",
"name": "Ministral 3 14B Instruct 2512",
"description": "Compact Mistral VLM for chat and instruction-based workloads",
"family": "ministral",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/mistralai/ministral-14b-instruct-2512\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/ministral-14b-instruct-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-small-4-119b-2603": {
"id": "mistralai/mistral-small-4-119b-2603",
"name": "mistral-small-4-119b-2603",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/mistralai/mistral-small-4-119b-2603\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-small-4-119b-2603\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mixtral-8x7b-instruct": {
"id": "mistralai/mixtral-8x7b-instruct",
"name": "Mistral: Mixtral 8x7B Instruct",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2023-12-10",
"last_updated": "2026-03-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/mistralai/mixtral-8x7b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mixtral-8x7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-medium-3.5-128b": {
"id": "mistralai/mistral-medium-3.5-128b",
"name": "Mistral Medium 3.5",
"description": "Balanced Mistral model for enterprise assistants, multilingual work, and tools",
"family": "mistral-medium",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-29",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/mistralai/mistral-medium-3.5-128b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-medium-3.5-128b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-7b-instruct-v0.3": {
"id": "mistralai/mistral-7b-instruct-v0.3",
"name": "Mistral-7B-Instruct-v0.3",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-01",
"last_updated": "2025-04-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/mistralai/mistral-7b-instruct-v0.3\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-7b-instruct-v0.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/mistral-medium-3-instruct": {
"id": "mistralai/mistral-medium-3-instruct",
"name": "Mistral Medium 3",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-medium",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"input": 131072,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/mistralai/mistral-medium-3-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/mistral-medium-3-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/magistral-small-2506": {
"id": "mistralai/magistral-small-2506",
"name": "Magistral Small 2506",
"description": "Mistral reasoning model for transparent analysis, math, and complex decisions",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-09-25",
"last_updated": "2025-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"input": 32768,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/mistralai/magistral-small-2506\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/magistral-small-2506\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/streampetr": {
"id": "nvidia/streampetr",
"name": "streampetr",
"description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/streampetr\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/streampetr\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/llama-3.3-nemotron-super-49b-v1.5": {
"id": "nvidia/llama-3.3-nemotron-super-49b-v1.5",
"name": "Llama 3.3 Nemotron Super 49B v1.5",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-07-25",
"last_updated": "2025-07-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/llama-3.3-nemotron-super-49b-v1.5\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/llama-3.3-nemotron-super-49b-v1.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/usdcode": {
"id": "nvidia/usdcode",
"name": "usdcode",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-01-01",
"last_updated": "2026-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/usdcode\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/usdcode\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-nano-omni-30b-a3b-reasoning": {
"id": "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning",
"name": "Nemotron 3 Nano Omni",
"description": "Open Nemotron omni model combining reasoning with text, vision, and audio",
"family": "nemotron",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": -1,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-28",
"last_updated": "2026-04-28",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nemotron-3-nano-omni-30b-a3b-reasoning\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-nano-omni-30b-a3b-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/cosmos-transfer1-7b": {
"id": "nvidia/cosmos-transfer1-7b",
"name": "cosmos-transfer1-7b",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-06-13",
"last_updated": "2025-06-30",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"video"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/cosmos-transfer1-7b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/cosmos-transfer1-7b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-voicechat": {
"id": "nvidia/nemotron-voicechat",
"name": "nemotron-voicechat",
"description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
"family": "nemotron",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nemotron-voicechat\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-voicechat\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/studiovoice": {
"id": "nvidia/studiovoice",
"name": "studiovoice",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-10-03",
"last_updated": "2025-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/studiovoice\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/studiovoice\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-content-safety": {
"id": "nvidia/nemotron-3-content-safety",
"name": "nemotron-3-content-safety",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nemotron-3-content-safety\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-content-safety\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/cosmos-transfer2_5-2b": {
"id": "nvidia/cosmos-transfer2_5-2b",
"name": "cosmos-transfer2.5-2b",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-02-26",
"last_updated": "2026-02-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"video"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/cosmos-transfer2_5-2b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/cosmos-transfer2_5-2b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/bevformer": {
"id": "nvidia/bevformer",
"name": "bevformer",
"description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-03-18",
"last_updated": "2025-07-20",
"modalities": {
"input": [
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/bevformer\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/bevformer\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/llama-3.1-nemotron-nano-vl-8b-v1": {
"id": "nvidia/llama-3.1-nemotron-nano-vl-8b-v1",
"name": "Llama 3.1 Nemotron Nano VL 8B v1",
"description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
"family": "nemotron",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-04-10",
"last_updated": "2025-04-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/llama-3.1-nemotron-nano-vl-8b-v1\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/llama-3.1-nemotron-nano-vl-8b-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/magpie-tts-zeroshot": {
"id": "nvidia/magpie-tts-zeroshot",
"name": "magpie-tts-zeroshot",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-05-22",
"last_updated": "2025-06-12",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"audio"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/magpie-tts-zeroshot\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/magpie-tts-zeroshot\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-nano-12b-v2-vl": {
"id": "nvidia/nemotron-nano-12b-v2-vl",
"name": "Nemotron Nano 12B v2 VL",
"description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
"family": "nemotron",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-10-28",
"last_updated": "2025-10-28",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nemotron-nano-12b-v2-vl\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-nano-12b-v2-vl\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/sparsedrive": {
"id": "nvidia/sparsedrive",
"name": "sparsedrive",
"description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-03-18",
"last_updated": "2025-07-20",
"modalities": {
"input": [
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/sparsedrive\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/sparsedrive\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/llama-nemotron-embed-vl-1b-v2": {
"id": "nvidia/llama-nemotron-embed-vl-1b-v2",
"name": "llama-nemotron-embed-vl-1b-v2",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "nemotron",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-02-10",
"last_updated": "2026-02-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 2048
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/llama-nemotron-embed-vl-1b-v2\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/llama-nemotron-embed-vl-1b-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/synthetic-video-detector": {
"id": "nvidia/synthetic-video-detector",
"name": "synthetic-video-detector",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/synthetic-video-detector\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/synthetic-video-detector\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-super-120b-a12b": {
"id": "nvidia/nemotron-3-super-120b-a12b",
"name": "Nemotron 3 Super",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.2,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nemotron-3-super-120b-a12b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-super-120b-a12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/llama-nemotron-rerank-vl-1b-v2": {
"id": "nvidia/llama-nemotron-rerank-vl-1b-v2",
"name": "llama-nemotron-rerank-vl-1b-v2",
"description": "Reranking model for improving retrieval quality in search and recommendation systems",
"family": "nemotron",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-03-31",
"last_updated": "2026-03-31",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/llama-nemotron-rerank-vl-1b-v2\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/llama-nemotron-rerank-vl-1b-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/usdvalidate": {
"id": "nvidia/usdvalidate",
"name": "usdvalidate",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-07-24",
"last_updated": "2025-01-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/usdvalidate\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/usdvalidate\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/active-speaker-detection": {
"id": "nvidia/active-speaker-detection",
"name": "Active Speaker Detection",
"description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/active-speaker-detection\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/active-speaker-detection\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/llama-3.1-nemotron-ultra-253b-v1": {
"id": "nvidia/llama-3.1-nemotron-ultra-253b-v1",
"name": "Llama 3.1 Nemotron Ultra 253B",
"description": "Flagship Nemotron model for high-throughput reasoning and complex agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-04-07",
"last_updated": "2025-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/llama-3.1-nemotron-ultra-253b-v1\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/llama-3.1-nemotron-ultra-253b-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/llama-3_2-nemoretriever-300m-embed-v1": {
"id": "nvidia/llama-3_2-nemoretriever-300m-embed-v1",
"name": "llama-3_2-nemoretriever-300m-embed-v1",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-07-24",
"last_updated": "2025-07-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 2048
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/llama-3_2-nemoretriever-300m-embed-v1\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/llama-3_2-nemoretriever-300m-embed-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nv-embedcode-7b-v1": {
"id": "nvidia/nv-embedcode-7b-v1",
"name": "nv-embedcode-7b-v1",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-03-17",
"last_updated": "2025-05-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 2048
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nv-embedcode-7b-v1\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nv-embedcode-7b-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/llama-3.1-nemotron-safety-guard-8b-v3": {
"id": "nvidia/llama-3.1-nemotron-safety-guard-8b-v3",
"name": "llama-3.1-nemotron-safety-guard-8b-v3",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-10-28",
"last_updated": "2025-10-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/llama-3.1-nemotron-safety-guard-8b-v3\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/llama-3.1-nemotron-safety-guard-8b-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-mini-4b-instruct": {
"id": "nvidia/nemotron-mini-4b-instruct",
"name": "nemotron-mini-4b-instruct",
"description": "Compact Nemotron model for efficient reasoning and deployable AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-08-21",
"last_updated": "2024-08-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nemotron-mini-4b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-mini-4b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/cosmos-predict1-5b": {
"id": "nvidia/cosmos-predict1-5b",
"name": "cosmos-predict1-5b",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-03-18",
"last_updated": "2025-03-18",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"video"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/cosmos-predict1-5b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/cosmos-predict1-5b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-content-safety-reasoning-4b": {
"id": "nvidia/nemotron-content-safety-reasoning-4b",
"name": "nemotron-content-safety-reasoning-4b",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": false,
"release_date": "2026-01-22",
"last_updated": "2026-01-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nemotron-content-safety-reasoning-4b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-content-safety-reasoning-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/riva-translate-4b-instruct-v1.1": {
"id": "nvidia/riva-translate-4b-instruct-v1.1",
"name": "riva-translate-4b-instruct-v1_1",
"description": "Translation model for multilingual conversion, localization, and cross-language workflows",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-12-12",
"last_updated": "2025-12-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/riva-translate-4b-instruct-v1.1\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/riva-translate-4b-instruct-v1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nvidia-nemotron-nano-9b-v2": {
"id": "nvidia/nvidia-nemotron-nano-9b-v2",
"name": "nvidia-nemotron-nano-9b-v2",
"description": "Compact Nemotron model for efficient reasoning and deployable AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2025-08-18",
"last_updated": "2025-08-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nvidia-nemotron-nano-9b-v2\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nvidia-nemotron-nano-9b-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/gliner-pii": {
"id": "nvidia/gliner-pii",
"name": "gliner-pii",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/gliner-pii\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/gliner-pii\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/llama-3.1-nemotron-nano-8b-v1": {
"id": "nvidia/llama-3.1-nemotron-nano-8b-v1",
"name": "Llama 3.1 Nemotron Nano 8B v1",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-03-18",
"last_updated": "2025-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/llama-3.1-nemotron-nano-8b-v1\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/llama-3.1-nemotron-nano-8b-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nv-embed-v1": {
"id": "nvidia/nv-embed-v1",
"name": "nv-embed-v1",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-06-07",
"last_updated": "2025-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 2048
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nv-embed-v1\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nv-embed-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/rerank-qa-mistral-4b": {
"id": "nvidia/rerank-qa-mistral-4b",
"name": "rerank-qa-mistral-4b",
"description": "Reranking model for improving retrieval quality in search and recommendation systems",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-03-17",
"last_updated": "2025-01-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/rerank-qa-mistral-4b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/rerank-qa-mistral-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-ultra-550b-a55b": {
"id": "nvidia/nemotron-3-ultra-550b-a55b",
"name": "Nemotron 3 Ultra 550B A55B",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 2.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nemotron-3-ultra-550b-a55b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-ultra-550b-a55b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3.5-lightning-30b-a3b": {
"id": "nvidia/nemotron-3.5-lightning-30b-a3b",
"name": "Nemotron 3.5 Lightning 30B A3B",
"description": "Fast NVIDIA Nemotron MoE for reliable agentic tasks across enterprise workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-11",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nemotron-3.5-lightning-30b-a3b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3.5-lightning-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/llama-3.3-nemotron-super-49b-v1": {
"id": "nvidia/llama-3.3-nemotron-super-49b-v1",
"name": "Llama 3.3 Nemotron Super 49B v1",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-04-07",
"last_updated": "2025-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/llama-3.3-nemotron-super-49b-v1\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/llama-3.3-nemotron-super-49b-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-nano-30b-a3b": {
"id": "nvidia/nemotron-3-nano-30b-a3b",
"name": "nemotron-3-nano-30b-a3b",
"description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2024-12",
"last_updated": "2024-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/nemotron-3-nano-30b-a3b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-nano-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/llama-3.1-nemotron-70b-instruct": {
"id": "nvidia/llama-3.1-nemotron-70b-instruct",
"name": "Llama 3.1 Nemotron 70B Instruct",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-04-15",
"last_updated": "2025-04-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/llama-3.1-nemotron-70b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/llama-3.1-nemotron-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/cosmos-reason2-8b": {
"id": "nvidia/cosmos-reason2-8b",
"name": "Cosmos Reason2 8B",
"description": "Vision language model for physical-world understanding with structured reasoning on video and images",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/nvidia/cosmos-reason2-8b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/cosmos-reason2-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-4b-it": {
"id": "google/gemma-3-4b-it",
"name": "Gemma 3 4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-03-12",
"last_updated": "2025-03-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/google/gemma-3-4b-it\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3-4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3n-e2b-it": {
"id": "google/gemma-3n-e2b-it",
"name": "Gemma 3n E2b It",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2025-06-12",
"last_updated": "2025-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/google/gemma-3n-e2b-it\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3n-e2b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/google-paligemma": {
"id": "google/google-paligemma",
"name": "paligemma",
"description": "Gemini multimodal model for text, image, audio, video, and document tasks",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-05-14",
"last_updated": "2024-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/google/google-paligemma\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"google/google-paligemma\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-2-2b-it": {
"id": "google/gemma-2-2b-it",
"name": "Gemma 2 2b It",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-07-16",
"last_updated": "2024-07-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/google/gemma-2-2b-it\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-2-2b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31b-it": {
"id": "google/gemma-4-31b-it",
"name": "Gemma-4-31B-IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/google/gemma-4-31b-it\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3n-e4b-it": {
"id": "google/gemma-3n-e4b-it",
"name": "Gemma 3n E4b It",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2025-06-03",
"last_updated": "2025-06-03",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/google/gemma-3n-e4b-it\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3n-e4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-12b-it": {
"id": "google/gemma-3-12b-it",
"name": "Gemma 3 12B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-03-12",
"last_updated": "2025-03-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/google/gemma-3-12b-it\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3-12b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/inkling": {
"id": "thinkingmachines/inkling",
"name": "Inkling",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/thinkingmachines/inkling\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-3.1-8b-instruct": {
"id": "meta/llama-3.1-8b-instruct",
"name": "Llama 3.1 8B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2025-01-01",
"last_updated": "2025-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/llama-3.1-8b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-3.1-8b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-guard-4-12b": {
"id": "meta/llama-guard-4-12b",
"name": "Llama Guard 4 12B",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-04-05",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/llama-guard-4-12b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-guard-4-12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-3.2-90b-vision-instruct": {
"id": "meta/llama-3.2-90b-vision-instruct",
"name": "Llama-3.2-90B-Vision-Instruct",
"description": "Open Llama multimodal model for image understanding and text reasoning",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-09-25",
"last_updated": "2024-09-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/llama-3.2-90b-vision-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-3.2-90b-vision-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-3.2-3b-instruct": {
"id": "meta/llama-3.2-3b-instruct",
"name": "Llama 3.2 3B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2024-09-18",
"last_updated": "2024-09-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/llama-3.2-3b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-3.2-3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-3.2-1b-instruct": {
"id": "meta/llama-3.2-1b-instruct",
"name": "Llama 3.2 1b Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-09-18",
"last_updated": "2024-09-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/llama-3.2-1b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-3.2-1b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/esmfold": {
"id": "meta/esmfold",
"name": "esmfold",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-03-15",
"last_updated": "2025-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/esmfold\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/esmfold\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-4-maverick-17b-128e-instruct": {
"id": "meta/llama-4-maverick-17b-128e-instruct",
"name": "Llama 4 Maverick 17b 128e Instruct",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-02",
"release_date": "2025-04-01",
"last_updated": "2025-04-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/llama-4-maverick-17b-128e-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-4-maverick-17b-128e-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-glimmer-30b": {
"id": "meta/muse-glimmer-30b",
"name": "Muse Glimmer 30B",
"description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2026-01-04",
"release_date": "2026-08-10",
"last_updated": "2026-08-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/muse-glimmer-30b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-glimmer-30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/esm2-650m": {
"id": "meta/esm2-650m",
"name": "esm2-650m",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-08-29",
"last_updated": "2025-03-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/esm2-650m\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/esm2-650m\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-3.2-11b-vision-instruct": {
"id": "meta/llama-3.2-11b-vision-instruct",
"name": "Llama 3.2 11b Vision Instruct",
"description": "Open Llama multimodal model for image understanding and text reasoning",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-09-18",
"last_updated": "2024-09-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/llama-3.2-11b-vision-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-3.2-11b-vision-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-3.1-70b-instruct": {
"id": "meta/llama-3.1-70b-instruct",
"name": "Llama 3.1 70b Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-07-16",
"last_updated": "2024-07-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/llama-3.1-70b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-3.1-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-3.3-70b-instruct": {
"id": "meta/llama-3.3-70b-instruct",
"name": "Llama 3.3 70b Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-11-26",
"last_updated": "2024-11-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/meta/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seed-oss-36b-instruct": {
"id": "bytedance/seed-oss-36b-instruct",
"name": "ByteDance-Seed/Seed-OSS-36B-Instruct",
"description": "Tool-capable chat model for instruction following and agentic application workflows",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-04",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/bytedance/seed-oss-36b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seed-oss-36b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sarvamai/sarvam-m": {
"id": "sarvamai/sarvam-m",
"name": "sarvam-m",
"description": "Efficient Indian-language reasoning model for chat, coding, and multilingual work",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-07-25",
"last_updated": "2025-07-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/sarvamai/sarvam-m\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"sarvamai/sarvam-m\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"microsoft/phi-4-multimodal-instruct": {
"id": "microsoft/phi-4-multimodal-instruct",
"name": "Phi 4 Multimodal",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"release_date": "2025-07-26",
"last_updated": "2025-07-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 128000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/microsoft/phi-4-multimodal-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"microsoft/phi-4-multimodal-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"microsoft/phi-4-mini-instruct": {
"id": "microsoft/phi-4-mini-instruct",
"name": "Phi-4-Mini",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "phi",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2024-12-01",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/microsoft/phi-4-mini-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"microsoft/phi-4-mini-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimaxai/minimax-m2.7": {
"id": "minimaxai/minimax-m2.7",
"name": "MiniMax-M2.7",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-04-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/minimaxai/minimax-m2.7\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"minimaxai/minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimaxai/minimax-m3": {
"id": "minimaxai/minimax-m3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/minimaxai/minimax-m3\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"minimaxai/minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baai/bge-m3": {
"id": "baai/bge-m3",
"name": "BGE M3",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "bge",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-01-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 1024
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/baai/bge-m3\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"baai/bge-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"abacusai/dracarys-llama-3.1-70b-instruct": {
"id": "abacusai/dracarys-llama-3.1-70b-instruct",
"name": "dracarys-llama-3.1-70b-instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-09-11",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/abacusai/dracarys-llama-3.1-70b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"abacusai/dracarys-llama-3.1-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/openai/gpt-oss-20b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/whisper-large-v3": {
"id": "openai/whisper-large-v3",
"name": "Whisper Large v3",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "whisper",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2023-09",
"release_date": "2023-09-01",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 0,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/openai/whisper-large-v3\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"openai/whisper-large-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT-OSS-120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08",
"release_date": "2025-08-04",
"last_updated": "2025-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/openai/gpt-oss-120b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.6": {
"id": "moonshotai/kimi-k2.6",
"name": "Kimi K2.6",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"status": "deprecated",
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/moonshotai/kimi-k2.6\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-instruct-0905": {
"id": "moonshotai/kimi-k2-instruct-0905",
"name": "Kimi K2 0905",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-09-05",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"status": "deprecated",
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/moonshotai/kimi-k2-instruct-0905\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-instruct-0905\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k3": {
"id": "moonshotai/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/moonshotai/kimi-k3\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"upstage/solar-10.7b-instruct": {
"id": "upstage/solar-10.7b-instruct",
"name": "solar-10.7b-instruct",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-06-05",
"last_updated": "2025-04-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/upstage/solar-10.7b-instruct\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"upstage/solar-10.7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.2": {
"id": "z-ai/glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/z-ai/glm-5.2\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"black-forest-labs/flux_1-kontext-dev": {
"id": "black-forest-labs/flux_1-kontext-dev",
"name": "FLUX.1-Kontext-dev",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-08-12",
"last_updated": "2025-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 40960
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/black-forest-labs/flux_1-kontext-dev\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"black-forest-labs/flux_1-kontext-dev\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"black-forest-labs/flux_1-schnell": {
"id": "black-forest-labs/flux_1-schnell",
"name": "FLUX.1-schnell",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": false,
"knowledge": "2024-07",
"release_date": "2024-08-01",
"last_updated": "2026-02-04",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": true,
"limit": {
"context": 77,
"input": 77,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/black-forest-labs/flux_1-schnell\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"black-forest-labs/flux_1-schnell\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"black-forest-labs/flux_2-klein-4b": {
"id": "black-forest-labs/flux_2-klein-4b",
"name": "FLUX.2 Klein 4B",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2026-01-14",
"last_updated": "2026-01-31",
"modalities": {
"input": [
"image",
"text"
],
"output": [
"image"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 40960
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/black-forest-labs/flux_2-klein-4b\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"black-forest-labs/flux_2-klein-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"black-forest-labs/flux.1-dev": {
"id": "black-forest-labs/flux.1-dev",
"name": "FLUX.1-dev",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2024-08-01",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 4096,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"nvidia/black-forest-labs/flux.1-dev\", apiKey: processEnvironment[\"NVIDIA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://integrate.api.nvidia.com/v1\")!,\n apiKey: processEnvironment[\"NVIDIA_API_KEY\"]\n)\nlet session = provider.model(\"black-forest-labs/flux.1-dev\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"jiekou": {
"id": "jiekou",
"name": "Jiekou.AI",
"baseURL": "https://api.jiekou.ai/openai",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"JIEKOU_API_KEY"
],
"doc": "https://docs.jiekou.ai/docs/support/quickstart?utm_source=github_models.dev",
"modelCount": 61,
"models": {
"gpt-5-nano": {
"id": "gpt-5-nano",
"name": "gpt-5-nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.045,
"output": 0.36
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5-nano\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-1-fast-non-reasoning": {
"id": "grok-4-1-fast-non-reasoning",
"name": "grok-4-1-fast-non-reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 0.18,
"output": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/grok-4-1-fast-non-reasoning\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-1-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-codex": {
"id": "gpt-5-codex",
"name": "gpt-5-codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.125,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5-codex\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-pro": {
"id": "gpt-5-pro",
"name": "gpt-5-pro",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 272000
},
"cost": {
"input": 13.5,
"output": 108
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5-pro\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex-mini": {
"id": "gpt-5.1-codex-mini",
"name": "gpt-5.1-codex-mini",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.225,
"output": 1.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5.1-codex-mini\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex": {
"id": "gpt-5.1-codex",
"name": "gpt-5.1-codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.125,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5.1-codex\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-code-fast-1": {
"id": "grok-code-fast-1",
"name": "grok-code-fast-1",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.18,
"output": 1.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/grok-code-fast-1\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"grok-code-fast-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2-codex": {
"id": "gpt-5.2-codex",
"name": "gpt-5.2-codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5.2-codex\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.2-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro-preview-06-05": {
"id": "gemini-2.5-pro-preview-06-05",
"name": "gemini-2.5-pro-preview-06-05",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 200000
},
"cost": {
"input": 1.125,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gemini-2.5-pro-preview-06-05\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro-preview-06-05\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite": {
"id": "gemini-2.5-flash-lite",
"name": "gemini-2.5-flash-lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.09,
"output": 0.36
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2-pro": {
"id": "gpt-5.2-pro",
"name": "gpt-5.2-pro",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 18.9,
"output": 151.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5.2-pro\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-pro-preview": {
"id": "gemini-3-pro-preview",
"name": "gemini-3-pro-preview",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.8,
"output": 10.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gemini-3-pro-preview\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1-20250805": {
"id": "claude-opus-4-1-20250805",
"name": "claude-opus-4-1-20250805",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 13.5,
"output": 67.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/claude-opus-4-1-20250805\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-1-20250805\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-chat-latest": {
"id": "gpt-5-chat-latest",
"name": "gpt-5-chat-latest",
"description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.125,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5-chat-latest\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-chat-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-1-fast-reasoning": {
"id": "grok-4-1-fast-reasoning",
"name": "grok-4-1-fast-reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 0.18,
"output": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/grok-4-1-fast-reasoning\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-1-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite-preview-06-17": {
"id": "gemini-2.5-flash-lite-preview-06-17",
"name": "gemini-2.5-flash-lite-preview-06-17",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"video",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.09,
"output": 0.36
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gemini-2.5-flash-lite-preview-06-17\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite-preview-06-17\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-20250514": {
"id": "claude-opus-4-20250514",
"name": "claude-opus-4-20250514",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 13.5,
"output": 67.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/claude-opus-4-20250514\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-20250514\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1": {
"id": "gpt-5.1",
"name": "gpt-5.1",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02",
"last_updated": "2026-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.125,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5.1\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex-max": {
"id": "gpt-5.1-codex-max",
"name": "gpt-5.1-codex-max",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.125,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5.1-codex-max\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-codex-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite-preview-09-2025": {
"id": "gemini-2.5-flash-lite-preview-09-2025",
"name": "gemini-2.5-flash-lite-preview-09-2025",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.09,
"output": 0.36
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gemini-2.5-flash-lite-preview-09-2025\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite-preview-09-2025\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "claude-opus-4-6",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024,
"max": 127999
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02",
"last_updated": "2026-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/claude-opus-4-6\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-fast-reasoning": {
"id": "grok-4-fast-reasoning",
"name": "grok-4-fast-reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 0.18,
"output": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/grok-4-fast-reasoning\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5-20250929": {
"id": "claude-sonnet-4-5-20250929",
"name": "claude-sonnet-4-5-20250929",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 2.7,
"output": 13.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/claude-sonnet-4-5-20250929\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5-20250929\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5-20251001": {
"id": "claude-haiku-4-5-20251001",
"name": "claude-haiku-4-5-20251001",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 20000,
"output": 64000
},
"cost": {
"input": 0.9,
"output": 4.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/claude-haiku-4-5-20251001\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5-20251001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-0709": {
"id": "grok-4-0709",
"name": "grok-4-0709",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 8192
},
"cost": {
"input": 2.7,
"output": 13.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/grok-4-0709\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-0709\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-preview-05-20": {
"id": "gemini-2.5-flash-preview-05-20",
"name": "gemini-2.5-flash-preview-05-20",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 200000
},
"cost": {
"input": 0.135,
"output": 3.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gemini-2.5-flash-preview-05-20\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-preview-05-20\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-flash-preview": {
"id": "gemini-3-flash-preview",
"name": "gemini-3-flash-preview",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gemini-3-flash-preview\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-mini": {
"id": "gpt-5-mini",
"name": "gpt-5-mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.225,
"output": 1.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5-mini\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "gemini-2.5-pro",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 1.125,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gemini-2.5-pro\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2": {
"id": "gpt-5.2",
"name": "gpt-5.2",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.575,
"output": 12.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gpt-5.2\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-20250514": {
"id": "claude-sonnet-4-20250514",
"name": "claude-sonnet-4-20250514",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 2.7,
"output": 13.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/claude-sonnet-4-20250514\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-20250514\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "gemini-2.5-flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65535
},
"cost": {
"input": 0.27,
"output": 2.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/gemini-2.5-flash\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-fast-non-reasoning": {
"id": "grok-4-fast-non-reasoning",
"name": "grok-4-fast-non-reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 0.18,
"output": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/grok-4-fast-non-reasoning\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o4-mini": {
"id": "o4-mini",
"name": "o4-mini",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/o4-mini\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"o4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o3-mini": {
"id": "o3-mini",
"name": "o3-mini",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 1.1,
"output": 4.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/o3-mini\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5-20251101": {
"id": "claude-opus-4-5-20251101",
"name": "claude-opus-4-5-20251101",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 65536
},
"cost": {
"input": 4.5,
"output": 22.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/claude-opus-4-5-20251101\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-5-20251101\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o3": {
"id": "o3",
"name": "o3",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 10,
"output": 40
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/o3\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-30b-a3b-fp8": {
"id": "qwen/qwen3-30b-a3b-fp8",
"name": "Qwen3 30B A3B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 20000
},
"cost": {
"input": 0.09,
"output": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/qwen/qwen3-30b-a3b-fp8\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-30b-a3b-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-next-80b-a3b-thinking": {
"id": "qwen/qwen3-next-80b-a3b-thinking",
"name": "Qwen3 Next 80B A3B Thinking",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/qwen/qwen3-next-80b-a3b-thinking\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-next-80b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b-thinking-2507": {
"id": "qwen/qwen3-235b-a22b-thinking-2507",
"name": "Qwen3 235B A22b Thinking 2507",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/qwen/qwen3-235b-a22b-thinking-2507\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b-thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-next-80b-a3b-instruct": {
"id": "qwen/qwen3-next-80b-a3b-instruct",
"name": "Qwen3 Next 80B A3B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/qwen/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b-fp8": {
"id": "qwen/qwen3-235b-a22b-fp8",
"name": "Qwen3 235B A22B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 20000
},
"cost": {
"input": 0.2,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/qwen/qwen3-235b-a22b-fp8\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-next": {
"id": "qwen/qwen3-coder-next",
"name": "qwen/qwen3-coder-next",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02",
"last_updated": "2026-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.2,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/qwen/qwen3-coder-next\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-32b-fp8": {
"id": "qwen/qwen3-32b-fp8",
"name": "Qwen3 32B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 20000
},
"cost": {
"input": 0.1,
"output": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/qwen/qwen3-32b-fp8\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-32b-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b-instruct-2507": {
"id": "qwen/qwen3-235b-a22b-instruct-2507",
"name": "Qwen3 235B A22B Instruct 2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/qwen/qwen3-235b-a22b-instruct-2507\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-480b-a35b-instruct": {
"id": "qwen/qwen3-coder-480b-a35b-instruct",
"name": "Qwen3 Coder 480B A35B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.29,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/qwen/qwen3-coder-480b-a35b-instruct\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-480b-a35b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/ernie-4.5-300b-a47b-paddle": {
"id": "baidu/ernie-4.5-300b-a47b-paddle",
"name": "ERNIE 4.5 300B A47B",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"family": "ernie",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 123000,
"output": 12000
},
"cost": {
"input": 0.28,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/baidu/ernie-4.5-300b-a47b-paddle\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"baidu/ernie-4.5-300b-a47b-paddle\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/ernie-4.5-vl-424b-a47b": {
"id": "baidu/ernie-4.5-vl-424b-a47b",
"name": "ERNIE 4.5 VL 424B A47B",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "ernie",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 123000,
"output": 16000
},
"cost": {
"input": 0.42,
"output": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/baidu/ernie-4.5-vl-424b-a47b\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"baidu/ernie-4.5-vl-424b-a47b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.1": {
"id": "minimax/minimax-m2.1",
"name": "Minimax M2.1",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 131071
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/minimax/minimax-m2.1\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.7": {
"id": "zai-org/glm-4.7",
"name": "GLM-4.7",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/zai-org/glm-4.7\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.5": {
"id": "zai-org/glm-4.5",
"name": "GLM-4.5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0.6,
"output": 2.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/zai-org/glm-4.5\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.5v": {
"id": "zai-org/glm-4.5v",
"name": "GLM 4.5V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glmv",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 16384
},
"cost": {
"input": 0.6,
"output": 1.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/zai-org/glm-4.5v\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-4.5v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-4.7-flash": {
"id": "zai-org/glm-4.7-flash",
"name": "GLM-4.7-Flash",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.07,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/zai-org/glm-4.7-flash\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-4.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimaxai/minimax-m1-80k": {
"id": "minimaxai/minimax-m1-80k",
"name": "MiniMax M1",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 40000
},
"cost": {
"input": 0.55,
"output": 2.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/minimaxai/minimax-m1-80k\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"minimaxai/minimax-m1-80k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-r1-0528": {
"id": "deepseek/deepseek-r1-0528",
"name": "DeepSeek R1 0528",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 32768
},
"cost": {
"input": 0.7,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/deepseek/deepseek-r1-0528\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-r1-0528\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3-0324": {
"id": "deepseek/deepseek-v3-0324",
"name": "DeepSeek V3 0324",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 0.28,
"output": 1.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/deepseek/deepseek-v3-0324\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3-0324\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.1": {
"id": "deepseek/deepseek-v3.1",
"name": "DeepSeek V3.1",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 32767
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 32768
},
"cost": {
"input": 0.27,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/deepseek/deepseek-v3.1\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomimimo/mimo-v2-flash": {
"id": "xiaomimimo/mimo-v2-flash",
"name": "XiaomiMiMo/MiMo-V2-Flash",
"description": "MiMo flash model for fast multimodal assistance and agent workflows",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/xiaomimimo/mimo-v2-flash\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"xiaomimimo/mimo-v2-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-0905": {
"id": "moonshotai/kimi-k2-0905",
"name": "Kimi K2 0905",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/moonshotai/kimi-k2-0905\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-0905\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-instruct": {
"id": "moonshotai/kimi-k2-instruct",
"name": "Kimi K2 Instruct",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.57,
"output": 2.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/moonshotai/kimi-k2-instruct\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.5": {
"id": "moonshotai/kimi-k2.5",
"name": "Kimi K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 262143
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jiekou/moonshotai/kimi-k2.5\", apiKey: processEnvironment[\"JIEKOU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jiekou.ai/openai\")!,\n apiKey: processEnvironment[\"JIEKOU_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"frogbot": {
"id": "frogbot",
"name": "FrogBot",
"baseURL": "https://app.frogbot.ai/api/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"FROGBOT_API_KEY"
],
"doc": "https://docs.frogbot.ai",
"modelCount": 26,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-02-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/claude-sonnet-4-6\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-1-fast-non-reasoning": {
"id": "grok-4-1-fast-non-reasoning",
"name": "Grok 4.1 Fast (Non-Reasoning)",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-11",
"release_date": "2025-11-25",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/grok-4-1-fast-non-reasoning\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-1-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-4-nano": {
"id": "gpt-5-4-nano",
"name": "GPT-5.4 Nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/gpt-5-4-nano\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-3-6-plus": {
"id": "qwen-3-6-plus",
"name": "Qwen 3.6 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-03",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/qwen-3-6-plus\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"qwen-3-6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-code-fast-1": {
"id": "grok-code-fast-1",
"name": "Grok 4.1 Fast (Reasoning)",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2023-10",
"release_date": "2025-08-28",
"last_updated": "2025-08-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.5,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/grok-code-fast-1\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"grok-code-fast-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-3-codex": {
"id": "gpt-5-3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/gpt-5-3-codex\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-20b": {
"id": "gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "1970-01-01",
"last_updated": "1970-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.07,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/gpt-oss-20b\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-1-fast-reasoning": {
"id": "grok-4-1-fast-reasoning",
"name": "Grok 4.1 Fast (Reasoning)",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-11",
"release_date": "2025-11-25",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/grok-4-1-fast-reasoning\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-1-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-6": {
"id": "kimi-k2-6",
"name": "Kimi-K2.6",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "1970-01-01",
"last_updated": "1970-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/kimi-k2-6\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/claude-opus-4-6\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2-5": {
"id": "minimax-m2-5",
"name": "MiniMax-M2.5",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2025-01-15",
"last_updated": "2025-02-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 192000,
"output": 8192
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/minimax-m2-5\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o": {
"id": "gpt-4o",
"name": "GPT-4o",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/gpt-4o\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7": {
"id": "claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/claude-opus-4-7\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-1-pro-preview": {
"id": "gemini-3-1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-01",
"release_date": "2026-02-18",
"last_updated": "2026-02-18",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/gemini-3-1-pro-preview\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-5": {
"id": "gpt-5-5",
"name": "GPT-5.5",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 272000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/gpt-5-5\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-glm-5-1": {
"id": "zai-glm-5-1",
"name": "Z.AI GLM-5.1",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-01-20",
"last_updated": "2025-02-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 198000,
"output": 8192
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/zai-glm-5-1\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"zai-glm-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-4-mini": {
"id": "gpt-5-4-mini",
"name": "GPT-5.4 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/gpt-5-4-mini\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-3": {
"id": "grok-4-3",
"name": "Grok 4.3",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-11",
"release_date": "2026-04-30",
"last_updated": "2026-04-30",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/grok-4-3\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5": {
"id": "claude-haiku-4-5",
"name": "Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/claude-haiku-4-5\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "Kimi-K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "1970-01-01",
"last_updated": "1970-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/kimi-k2.5\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-flash-preview": {
"id": "gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/gemini-3-flash-preview\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek v4 Pro",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2026-01",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 1.74,
"output": 3.48,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/deepseek-v4-pro\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "1970-01-01",
"last_updated": "1970-01-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/gpt-oss-120b\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-03-20",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.31
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/gemini-2.5-pro\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2-7": {
"id": "minimax-m2-7",
"name": "MiniMax-M2.7",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 192000,
"output": 8192
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/minimax-m2-7\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-07-17",
"last_updated": "2025-07-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"frogbot/gemini-2.5-flash\", apiKey: processEnvironment[\"FROGBOT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://app.frogbot.ai/api/v1\")!,\n apiKey: processEnvironment[\"FROGBOT_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"ovhcloud": {
"id": "ovhcloud",
"name": "OVHcloud AI Endpoints",
"baseURL": "https://oai.endpoints.kepler.ai.cloud.ovh.net/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"OVHCLOUD_API_KEY"
],
"doc": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/catalog//",
"modelCount": 15,
"models": {
"qwen3guard-gen-0.6b": {
"id": "qwen3guard-gen-0.6b",
"name": "Qwen3Guard-Gen-0.6B",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-01-22",
"last_updated": "2026-01-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/qwen3guard-gen-0.6b\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"qwen3guard-gen-0.6b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-9b": {
"id": "qwen3.5-9b",
"name": "Qwen3.5-9B",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.12,
"output": 0.18
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/qwen3.5-9b\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-27b": {
"id": "qwen3.8-27b",
"name": "Qwen3.8-27B",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/qwen3.8-27b\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-32b": {
"id": "qwen3-32b",
"name": "Qwen3-32B",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-16",
"last_updated": "2025-07-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.09,
"output": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/qwen3-32b\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-20b": {
"id": "gpt-oss-20b",
"name": "gpt-oss-20b",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2025-08-28",
"last_updated": "2025-08-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.18
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/gpt-oss-20b\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen2.5-vl-72b-instruct": {
"id": "qwen2.5-vl-72b-instruct",
"name": "Qwen2.5-VL-72B-Instruct",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2025-03-31",
"last_updated": "2025-03-31",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 1.01,
"output": 1.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/qwen2.5-vl-72b-instruct\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"qwen2.5-vl-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-30b-a3b-instruct": {
"id": "qwen3-coder-30b-a3b-instruct",
"name": "Qwen3-Coder-30B-A3B-Instruct",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-28",
"last_updated": "2025-10-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.07,
"output": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-397b-a17b": {
"id": "qwen3.5-397b-a17b",
"name": "Qwen3.5-397B-A17B",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-18",
"last_updated": "2026-05-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.71,
"output": 4.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-27b": {
"id": "qwen3.6-27b",
"name": "Qwen3.6-27B",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.47,
"output": 3.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/qwen3.6-27b\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-nemo-instruct-2407": {
"id": "mistral-nemo-instruct-2407",
"name": "Mistral-Nemo-Instruct-2407",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-11-20",
"last_updated": "2024-11-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.14,
"output": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/mistral-nemo-instruct-2407\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"mistral-nemo-instruct-2407\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-3.2-24b-instruct-2506": {
"id": "mistral-small-3.2-24b-instruct-2506",
"name": "Mistral-Small-3.2-24B-Instruct-2506",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-16",
"last_updated": "2025-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.31
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/mistral-small-3.2-24b-instruct-2506\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small-3.2-24b-instruct-2506\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-7b-instruct-v0.3": {
"id": "mistral-7b-instruct-v0.3",
"name": "Mistral-7B-Instruct-v0.3",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-01",
"last_updated": "2025-04-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.11,
"output": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/mistral-7b-instruct-v0.3\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"mistral-7b-instruct-v0.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3guard-gen-8b": {
"id": "qwen3guard-gen-8b",
"name": "Qwen3Guard-Gen-8B",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-01-22",
"last_updated": "2026-01-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/qwen3guard-gen-8b\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"qwen3guard-gen-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "gpt-oss-120b",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2025-08-28",
"last_updated": "2025-08-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.09,
"output": 0.47
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/gpt-oss-120b\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama-3_3-70b-instruct": {
"id": "meta-llama-3_3-70b-instruct",
"name": "Meta-Llama-3_3-70B-Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-01",
"last_updated": "2025-04-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.74,
"output": 0.74
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"ovhcloud/meta-llama-3_3-70b-instruct\", apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://oai.endpoints.kepler.ai.cloud.ovh.net/v1\")!,\n apiKey: processEnvironment[\"OVHCLOUD_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama-3_3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"xpersona": {
"id": "xpersona",
"name": "Xpersona",
"baseURL": "https://www.xpersona.co/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"XPERSONA_API_KEY"
],
"doc": "https://www.xpersona.co/docs",
"modelCount": 13,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.9,
"output": 5.55,
"reasoning": 5.55,
"cache_read": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/claude-sonnet-4-6\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-sol": {
"id": "gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 372000,
"output": 128000
},
"cost": {
"input": 1.5,
"output": 12,
"reasoning": 12,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/gpt-5.6-sol\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xpersona-gpt-5.5": {
"id": "xpersona-gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-30",
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 3,
"output": 18,
"reasoning": 18,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/xpersona-gpt-5.5\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"xpersona-gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 6,
"reasoning": 6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/gpt-5.4\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash": {
"id": "gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 1.55,
"output": 12.2,
"reasoning": 12.2,
"cache_read": 0.155
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/gemini-3.5-flash\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 3,
"output": 18.5,
"reasoning": 18.5,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/claude-fable-5\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xpersona-frieren-coder": {
"id": "xpersona-frieren-coder",
"name": "Xpersona Frieren 1",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-30",
"release_date": "2026-05-01",
"last_updated": "2026-05-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 1.5,
"output": 6,
"reasoning": 6,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/xpersona-frieren-coder\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"xpersona-frieren-coder\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-mini": {
"id": "gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 272000,
"output": 128000
},
"cost": {
"input": 0.375,
"output": 4,
"reasoning": 4,
"cache_read": 0.0375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/gpt-5.4-mini\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5": {
"id": "claude-haiku-4-5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.6,
"output": 3.7,
"reasoning": 3.7,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/claude-haiku-4-5\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 1.5,
"output": 9.25,
"reasoning": 9.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/claude-opus-4-8\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6": {
"id": "gpt-5.6",
"name": "GPT-5.6",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 372000,
"output": 128000
},
"cost": {
"input": 1.5,
"output": 12,
"reasoning": 12,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/gpt-5.6\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-terra": {
"id": "gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 372000,
"output": 128000
},
"cost": {
"input": 1.5,
"output": 2,
"reasoning": 2,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/gpt-5.6-terra\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"output": 128000
},
"cost": {
"input": 1.5,
"output": 12,
"reasoning": 12,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"xpersona/gpt-5.5\", apiKey: processEnvironment[\"XPERSONA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://www.xpersona.co/v1\")!,\n apiKey: processEnvironment[\"XPERSONA_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"anthropic": {
"id": "anthropic",
"name": "Anthropic",
"baseURL": "",
"npm": "@ai-sdk/anthropic",
"swiftDriver": "anthropicMessages",
"env": [
"ANTHROPIC_API_KEY"
],
"doc": "https://docs.anthropic.com/en/docs/about-claude/models",
"modelCount": 14,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-sonnet-4-6\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5": {
"id": "claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"provider": {
"body": {
"speed": "fast"
},
"headers": {
"anthropic-beta": "fast-mode-2026-02-01"
}
}
}
}
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-opus-5\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5": {
"id": "claude-opus-4-5",
"name": "Claude Opus 4.5 (latest)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-opus-4-5\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5-1": {
"id": "claude-fable-5-1",
"name": "Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-fable-5-1\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-04",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-opus-4-6\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5-20250929": {
"id": "claude-sonnet-4-5-20250929",
"name": "Claude Sonnet 4.5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-sonnet-4-5-20250929\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5-20250929\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7": {
"id": "claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-14",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-opus-4-7\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5-20251001": {
"id": "claude-haiku-4-5-20251001",
"name": "Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-haiku-4-5-20251001\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5-20251001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-07",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-fable-5\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5": {
"id": "claude-haiku-4-5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-haiku-4-5\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5": {
"id": "claude-sonnet-4-5",
"name": "Claude Sonnet 4.5 (latest)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-sonnet-4-5\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"provider": {
"body": {
"speed": "fast"
},
"headers": {
"anthropic-beta": "fast-mode-2026-02-01"
}
}
}
}
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-opus-4-8\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5": {
"id": "claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-29",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-sonnet-5\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5-20251101": {
"id": "claude-opus-4-5-20251101",
"name": "Claude Opus 4.5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"anthropic/claude-opus-4-5-20251101\", apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"ANTHROPIC_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-5-20251101\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"google": {
"id": "google",
"name": "Google",
"baseURL": "",
"npm": "@ai-sdk/google",
"swiftDriver": "geminiNative",
"env": [
"GOOGLE_API_KEY",
"GOOGLE_GENERATIVE_AI_API_KEY",
"GEMINI_API_KEY"
],
"doc": "https://ai.google.dev/gemini-api/docs/models",
"modelCount": 39,
"models": {
"gemma-4-26b-a4b-it": {
"id": "gemma-4-26b-a4b-it",
"name": "Gemma 4 26B A4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro-preview-customtools": {
"id": "gemini-3.1-pro-preview-customtools",
"name": "Gemini 3.1 Pro Preview Custom Tools",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.1-pro-preview-customtools\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-pro-preview-customtools\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite-image": {
"id": "gemini-3.1-flash-lite-image",
"name": "Nano Banana 2 Lite",
"description": "Fastest, most cost-efficient Gemini image model for high-volume 1K generation and editing",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 30
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.1-flash-lite-image\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"lyria-3-clip-preview": {
"id": "lyria-3-clip-preview",
"name": "Lyria 3 Clip Preview",
"description": "Music generation model for short 30-second clips, loops, and previews from text or image prompts",
"family": "lyria",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-25",
"last_updated": "2026-03-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/lyria-3-clip-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"lyria-3-clip-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-image": {
"id": "gemini-2.5-flash-image",
"name": "Nano Banana",
"description": "Nano Banana image model for fast generation, edits, and character-consistent assets",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2024-06",
"release_date": "2025-08-26",
"last_updated": "2025-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 30,
"cache_read": 0.075
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-2.5-flash-image\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deep-research-max-preview-04-2026": {
"id": "deep-research-max-preview-04-2026",
"name": "Deep Research Max Preview (Apr-21-2026)",
"description": "Maximum-comprehensiveness agentic researcher for multi-step investigation, synthesis, and cited reports",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/deep-research-max-preview-04-2026\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"deep-research-max-preview-04-2026\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-pro-image": {
"id": "gemini-3-pro-image",
"name": "Nano Banana Pro",
"description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 2,
"output": 120
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3-pro-image\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-pro-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro-preview": {
"id": "gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deep-research-preview-04-2026": {
"id": "deep-research-preview-04-2026",
"name": "Deep Research Preview (Apr-21-2026)",
"description": "Agentic model for autonomous multi-step research, synthesis, and cited reports",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/deep-research-preview-04-2026\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"deep-research-preview-04-2026\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite": {
"id": "gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash-Lite",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 512,
"max": 24576
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01,
"input_audio": 0.3
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-computer-use-preview-10-2025": {
"id": "gemini-2.5-computer-use-preview-10-2025",
"name": "Gemini 2.5 Computer Use Preview 10-2025",
"description": "Specialized Gemini 2.5 model for browser-control agents that automate UI tasks",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-10-07",
"last_updated": "2025-10-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"tiers": [
{
"input": 2.5,
"output": 15,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-2.5-computer-use-preview-10-2025\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-computer-use-preview-10-2025\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.6-flash": {
"id": "gemini-3.6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"input_audio": 0.75
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.6-flash\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite": {
"id": "gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"input_audio": 0.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-live-preview": {
"id": "gemini-3.1-flash-live-preview",
"name": "Gemini 3.1 Flash Live Preview",
"description": "High-quality, low-latency Live API model for real-time dialogue and voice-first AI applications",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-03-26",
"last_updated": "2026-03-26",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 4.5,
"input_audio": 3,
"output_audio": 12
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.1-flash-live-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-live-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro-preview-tts": {
"id": "gemini-2.5-pro-preview-tts",
"name": "Gemini 2.5 Pro Preview TTS",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "gemini-flash",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-05-01",
"last_updated": "2025-05-01",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 16384
},
"cost": {
"input": 1,
"output": 20
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-2.5-pro-preview-tts\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro-preview-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash": {
"id": "gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"input_audio": 1.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.5-flash\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"veo-3.1-generate-preview": {
"id": "veo-3.1-generate-preview",
"name": "Veo 3.1",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-10-15",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 8192
},
"status": "beta",
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/veo-3.1-generate-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"veo-3.1-generate-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite-preview": {
"id": "gemini-3.1-flash-lite-preview",
"name": "Gemini 3.1 Flash Lite Preview",
"description": "Legacy model retained for compatibility with older integrations",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"status": "deprecated",
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"input_audio": 0.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.1-flash-lite-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"veo-3.1-fast-generate-preview": {
"id": "veo-3.1-fast-generate-preview",
"name": "Veo 3.1 fast",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-10-15",
"last_updated": "2026-01-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 8192
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/veo-3.1-fast-generate-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"veo-3.1-fast-generate-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-preview-tts": {
"id": "gemini-2.5-flash-preview-tts",
"name": "Gemini 2.5 Flash Preview TTS",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "gemini-flash",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-05-01",
"last_updated": "2025-05-01",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 16384
},
"cost": {
"input": 0.5,
"output": 10
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-2.5-flash-preview-tts\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-preview-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-embedding-001": {
"id": "gemini-embedding-001",
"name": "Gemini Embedding 001",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "gemini",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2025-05",
"release_date": "2025-05-20",
"last_updated": "2025-05-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2048,
"output": 1
},
"cost": {
"input": 0.15,
"output": 0
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-embedding-001\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-embedding-001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-image": {
"id": "gemini-3.1-flash-image",
"name": "Nano Banana 2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 60
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.1-flash-image\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash-lite": {
"id": "gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-pro-image-preview": {
"id": "gemini-3-pro-image-preview",
"name": "Nano Banana Pro",
"description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-11-20",
"last_updated": "2025-11-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 2,
"output": 120
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3-pro-image-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-pro-image-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-tts-preview": {
"id": "gemini-3.1-flash-tts-preview",
"name": "Gemini 3.1 Flash TTS Preview",
"description": "Low-latency speech generation with steerable prompts and expressive audio tags",
"family": "gemini-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-15",
"last_updated": "2026-04-15",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 16384
},
"cost": {
"input": 1,
"output": 20
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.1-flash-tts-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-tts-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"veo-3.1-lite-generate-preview": {
"id": "veo-3.1-lite-generate-preview",
"name": "Veo 3.1 lite",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-03-31",
"last_updated": "2026-03-31",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 8192
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/veo-3.1-lite-generate-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"veo-3.1-lite-generate-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-flash-lite-latest": {
"id": "gemini-flash-lite-latest",
"name": "Gemini Flash-Lite Latest",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-flash-lite-latest\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-flash-lite-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it": {
"id": "gemma-4-31b-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemma-4-31b-it\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-embedding-2": {
"id": "gemini-embedding-2",
"name": "Gemini Embedding 2",
"description": "Multimodal embedding model mapping text, images, video, audio, and PDFs into a unified embedding space",
"family": "gemini",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"knowledge": "2025-11",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1
},
"cost": {
"input": 0.2,
"output": 0,
"input_audio": 6.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-embedding-2\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-embedding-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-flash-preview": {
"id": "gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"input_audio": 1
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3-flash-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.8-flash": {
"id": "gemini-3.8-flash",
"name": "Gemini 3.8 Flash",
"description": "Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"input_audio": 0.75
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.8-flash\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-live-translate-preview": {
"id": "gemini-3.5-live-translate-preview",
"name": "Gemini 3.5 Live Translate Preview",
"description": "Low-latency audio-to-audio model for real-time speech translation across 70+ languages",
"family": "gemini-pro",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"audio"
],
"output": [
"audio",
"text"
]
},
"open_weights": false,
"limit": {
"context": 16384,
"output": 32768
},
"cost": {
"input": 3.5,
"output": 21,
"input_audio": 3.5,
"output_audio": 21
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.5-live-translate-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-live-translate-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"lyria-3-pro-preview": {
"id": "lyria-3-pro-preview",
"name": "Lyria 3 Pro Preview",
"description": "Music generation model for full-length songs from text or images with vocals and structure",
"family": "lyria",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2026-03-25",
"last_updated": "2026-03-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/lyria-3-pro-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"lyria-3-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.7-flash": {
"id": "gemini-3.7-flash",
"name": "Gemini 3.7 Flash",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"input_audio": 0.75
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.7-flash\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-2.5-pro\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-flash-latest": {
"id": "gemini-flash-latest",
"name": "Gemini Flash Latest",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"input_audio": 0.75
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-flash-latest\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-flash-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-image-preview": {
"id": "gemini-3.1-flash-image-preview",
"name": "Nano Banana 2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-26",
"last_updated": "2026-02-26",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 60
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-3.1-flash-image-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-image-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-omni-flash-preview": {
"id": "gemini-omni-flash-preview",
"name": "Gemini Omni Flash Preview",
"description": "Video generation and editing model for fast, conversational text- and image-to-video workflows",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 17.5
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-omni-flash-preview\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-omni-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"input_audio": 1
},
"swiftDriver": "geminiNative",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"google/gemini-2.5-flash\", apiKey: processEnvironment[\"GOOGLE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"GOOGLE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"baseten": {
"id": "baseten",
"name": "Baseten",
"baseURL": "https://inference.baseten.co/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"BASETEN_API_KEY"
],
"doc": "https://docs.baseten.co/inference/model-apis/overview",
"modelCount": 23,
"models": {
"deepseek-ai/DeepSeek-V4-Flash-0731": {
"id": "deepseek-ai/DeepSeek-V4-Flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.13,
"output": 0.26,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/deepseek-ai/DeepSeek-V4-Flash-0731\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.1": {
"id": "deepseek-ai/DeepSeek-V3.1",
"name": "DeepSeek V3.1",
"description": "Legacy model retained for compatibility with older integrations",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-25",
"last_updated": "2025-08-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 164000,
"output": 131000
},
"status": "deprecated",
"cost": {
"input": 0.5,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/deepseek-ai/DeepSeek-V3.1\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4.1-Flash": {
"id": "deepseek-ai/DeepSeek-V4.1-Flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/deepseek-ai/DeepSeek-V4.1-Flash\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4.1-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro-0813": {
"id": "deepseek-ai/DeepSeek-V4-Pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 1.32,
"output": 3.96
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/deepseek-ai/DeepSeek-V4-Pro-0813\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro": {
"id": "deepseek-ai/DeepSeek-V4-Pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 1.74,
"output": 3.48,
"cache_read": 0.145
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/deepseek-ai/DeepSeek-V4-Pro\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B": {
"id": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B",
"name": "Nemotron Ultra",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202800,
"output": 202800
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Nemotron-120B-A12B": {
"id": "nvidia/Nemotron-120B-A12B",
"name": "Nemotron Super",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2026-02",
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202800,
"output": 202800
},
"cost": {
"input": 0.3,
"output": 0.75,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/nvidia/Nemotron-120B-A12B\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Nemotron-120B-A12B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.1": {
"id": "zai-org/GLM-5.1",
"name": "GLM 5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202800,
"output": 202800
},
"cost": {
"input": 1.3,
"output": 4.3,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/zai-org/GLM-5.1\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2-Fast": {
"id": "zai-org/GLM-5.2-Fast",
"name": "GLM 5.2 Fast",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 2.1,
"output": 6.6,
"cache_read": 0.21
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/zai-org/GLM-5.2-Fast\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2-Fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.3": {
"id": "zai-org/GLM-5.3",
"name": "GLM 5.3",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/zai-org/GLM-5.3\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2": {
"id": "zai-org/GLM-5.2",
"name": "GLM 5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/zai-org/GLM-5.2\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.7": {
"id": "zai-org/GLM-4.7",
"name": "GLM 4.7",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 200000
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/zai-org/GLM-4.7\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5": {
"id": "zai-org/GLM-5",
"name": "GLM 5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2026-01",
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202800,
"output": 202800
},
"cost": {
"input": 0.95,
"output": 3.15,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/zai-org/GLM-5\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.3-Fast": {
"id": "zai-org/GLM-5.3-Fast",
"name": "GLM 5.3 Fast",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 2.1,
"output": 6.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/zai-org/GLM-5.3-Fast\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.3-Fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.3-Flash": {
"id": "zai-org/GLM-5.3-Flash",
"name": "GLM 5.3 Flash",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/zai-org/GLM-5.3-Flash\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.3-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/inkling-small": {
"id": "thinkingmachines/inkling-small",
"name": "Inkling Small",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-07-30",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 1.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/thinkingmachines/inkling-small\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/inkling-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/inkling": {
"id": "thinkingmachines/inkling",
"name": "Inkling",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 32768
},
"cost": {
"input": 1,
"output": 4.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/thinkingmachines/inkling\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M2.5": {
"id": "MiniMaxAI/MiniMax-M2.5",
"name": "MiniMax-M2.5",
"description": "Legacy model retained for compatibility with older integrations",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2026-01",
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204000,
"output": 204000
},
"status": "deprecated",
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/MiniMaxAI/MiniMax-M2.5\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "OpenAI GPT 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-08",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128072,
"output": 128072
},
"cost": {
"input": 0.1,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/openai/gpt-oss-120b\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.5": {
"id": "moonshotai/Kimi-K2.5",
"name": "Kimi K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2026-01-30",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/moonshotai/Kimi-K2.5\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.7-Code": {
"id": "moonshotai/Kimi-K2.7-Code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/moonshotai/Kimi-K2.7-Code\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.7-Code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6": {
"id": "moonshotai/Kimi-K2.6",
"name": "Kimi K2.6",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K3": {
"id": "moonshotai/Kimi-K3",
"name": "Kimi K3",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 3,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"baseten/moonshotai/Kimi-K3\", apiKey: processEnvironment[\"BASETEN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.baseten.co/v1\")!,\n apiKey: processEnvironment[\"BASETEN_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"vercel": {
"id": "vercel",
"name": "Vercel AI Gateway",
"baseURL": "",
"npm": "@ai-sdk/gateway",
"swiftDriver": "openaiChat",
"env": [
"AI_GATEWAY_API_KEY"
],
"doc": "https://github.com/vercel/ai/tree/5eb85cc45a259553501f535b8ac79a77d0e79223/packages/gateway",
"modelCount": 372,
"models": {
"voyage/voyage-code-3": {
"id": "voyage/voyage-code-3",
"name": "voyage-code-3",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-12-04",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/voyage-code-3\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/voyage-code-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voyage/voyage-3.5": {
"id": "voyage/voyage-3.5",
"name": "voyage-3.5",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-05-20",
"last_updated": "2025-05-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/voyage-3.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/voyage-3.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voyage/voyage-3.5-lite": {
"id": "voyage/voyage-3.5-lite",
"name": "voyage-3.5-lite",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-05-20",
"last_updated": "2025-05-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/voyage-3.5-lite\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/voyage-3.5-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voyage/voyage-3-large": {
"id": "voyage/voyage-3-large",
"name": "voyage-3-large",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-01-07",
"last_updated": "2024-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/voyage-3-large\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/voyage-3-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voyage/voyage-code-2": {
"id": "voyage/voyage-code-2",
"name": "voyage-code-2",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-01-01",
"last_updated": "2024-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/voyage-code-2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/voyage-code-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voyage/voyage-4": {
"id": "voyage/voyage-4",
"name": "voyage-4",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-01-15",
"last_updated": "2026-03-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/voyage-4\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/voyage-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voyage/voyage-finance-2": {
"id": "voyage/voyage-finance-2",
"name": "voyage-finance-2",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-06-03",
"last_updated": "2024-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/voyage-finance-2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/voyage-finance-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voyage/rerank-2.5": {
"id": "voyage/rerank-2.5",
"name": "Voyage Rerank 2.5",
"description": "Reranking model for improving retrieval quality in search and recommendation systems",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-08-11",
"last_updated": "2025-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/rerank-2.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/rerank-2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voyage/voyage-law-2": {
"id": "voyage/voyage-law-2",
"name": "voyage-law-2",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-04-15",
"last_updated": "2024-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/voyage-law-2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/voyage-law-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voyage/voyage-4-large": {
"id": "voyage/voyage-4-large",
"name": "voyage-4-large",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-01-15",
"last_updated": "2026-03-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/voyage-4-large\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/voyage-4-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voyage/rerank-2.5-lite": {
"id": "voyage/rerank-2.5-lite",
"name": "Voyage Rerank 2.5 Lite",
"description": "Reranking model for improving retrieval quality in search and recommendation systems",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-08-11",
"last_updated": "2025-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/rerank-2.5-lite\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/rerank-2.5-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voyage/voyage-4-lite": {
"id": "voyage/voyage-4-lite",
"name": "voyage-4-lite",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "voyage",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-01-15",
"last_updated": "2026-03-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/voyage/voyage-4-lite\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"voyage/voyage-4-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"morph/morph-v3-fast": {
"id": "morph/morph-v3-fast",
"name": "Morph v3 Fast",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "morph",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-08-15",
"last_updated": "2024-08-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16000,
"output": 16000
},
"cost": {
"input": 0.8,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/morph/morph-v3-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"morph/morph-v3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"morph/morph-v3-large": {
"id": "morph/morph-v3-large",
"name": "Morph v3 Large",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "morph",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-08-15",
"last_updated": "2024-08-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 32000
},
"cost": {
"input": 0.9,
"output": 1.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/morph/morph-v3-large\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"morph/morph-v3-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-s-2.1-free": {
"id": "poolside/laguna-s-2.1-free",
"name": "Laguna S 2.1 Free",
"description": "Free provider route for experiments, demos, and cost-sensitive chat workloads",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/poolside/laguna-s-2.1-free\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-s-2.1-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-s-2.1": {
"id": "poolside/laguna-s-2.1",
"name": "Laguna S 2.1",
"description": "Agentic coding model from Poolside in the XS size class for local deployment",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/poolside/laguna-s-2.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-s-2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"klingai/kling-v2.5-turbo-i2v": {
"id": "klingai/kling-v2.5-turbo-i2v",
"name": "Kling v2.5 Turbo Image-to-Video",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/klingai/kling-v2.5-turbo-i2v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"klingai/kling-v2.5-turbo-i2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"klingai/kling-v3.0-motion-control": {
"id": "klingai/kling-v3.0-motion-control",
"name": "Kling v3.0 Motion Control",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-03-04",
"last_updated": "2026-03-04",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/klingai/kling-v3.0-motion-control\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"klingai/kling-v3.0-motion-control\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"klingai/kling-v2.6-t2v": {
"id": "klingai/kling-v2.6-t2v",
"name": "Kling v2.6 Text-to-Video",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-03",
"last_updated": "2025-12-21",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/klingai/kling-v2.6-t2v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"klingai/kling-v2.6-t2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"klingai/kling-v2.5-turbo-t2v": {
"id": "klingai/kling-v2.5-turbo-t2v",
"name": "Kling v2.5 Turbo Text-to-Video",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/klingai/kling-v2.5-turbo-t2v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"klingai/kling-v2.5-turbo-t2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"klingai/kling-v3.0-t2v": {
"id": "klingai/kling-v3.0-t2v",
"name": "Kling v3.0 Text-to-Video",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/klingai/kling-v3.0-t2v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"klingai/kling-v3.0-t2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"klingai/kling-v2.6-i2v": {
"id": "klingai/kling-v2.6-i2v",
"name": "Kling v2.6 Image-to-Video",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-03",
"last_updated": "2025-12-21",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/klingai/kling-v2.6-i2v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"klingai/kling-v2.6-i2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"klingai/kling-v2.6-motion-control": {
"id": "klingai/kling-v2.6-motion-control",
"name": "Kling v2.6 Motion Control",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-18",
"last_updated": "2025-12-21",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/klingai/kling-v2.6-motion-control\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"klingai/kling-v2.6-motion-control\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"klingai/kling-v3.0-i2v": {
"id": "klingai/kling-v3.0-i2v",
"name": "Kling v3.0 Image-to-Video",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "ling",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/klingai/kling-v3.0-i2v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"klingai/kling-v3.0-i2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kwaipilot/kat-coder-pro-v1": {
"id": "kwaipilot/kat-coder-pro-v1",
"name": "KAT-Coder-Pro V1",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"family": "kat-coder",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-11-09",
"last_updated": "2025-10-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/kwaipilot/kat-coder-pro-v1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kwaipilot/kat-coder-pro-v1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kwaipilot/kat-coder-pro-v2": {
"id": "kwaipilot/kat-coder-pro-v2",
"name": "Kat Coder Pro V2",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"family": "kat-coder",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-27",
"last_updated": "2026-03-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/kwaipilot/kat-coder-pro-v2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kwaipilot/kat-coder-pro-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kwaipilot/kat-coder-pro-v2.5": {
"id": "kwaipilot/kat-coder-pro-v2.5",
"name": "Kat Coder Pro V2.5",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"family": "kat-coder",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-10",
"last_updated": "2026-07-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 80000
},
"cost": {
"input": 0.74,
"output": 2.96,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/kwaipilot/kat-coder-pro-v2.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kwaipilot/kat-coder-pro-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kwaipilot/kat-coder-air-v2.5": {
"id": "kwaipilot/kat-coder-air-v2.5",
"name": "Kat Coder Air V2.5",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"family": "kat-coder",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-10",
"last_updated": "2026-07-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 80000
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/kwaipilot/kat-coder-air-v2.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kwaipilot/kat-coder-air-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun/step-3.7-flash": {
"id": "stepfun/step-3.7-flash",
"name": "Step 3.7 Flash",
"description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"cost": {
"input": 0.2,
"output": 1.15,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/stepfun/step-3.7-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"stepfun/step-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun/step-3.5-flash": {
"id": "stepfun/step-3.5-flash",
"name": "StepFun 3.5 Flash",
"description": "StepFun flash lane for quick multimodal reasoning and coding assistance",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-29",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262114,
"output": 262114
},
"cost": {
"input": 0.09,
"output": 0.3,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/stepfun/step-3.5-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"stepfun/step-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"interfaze/interfaze-beta": {
"id": "interfaze/interfaze-beta",
"name": "Interfaze Beta",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"release_date": "2025-10-07",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32000
},
"cost": {
"input": 1.5,
"output": 3.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/interfaze/interfaze-beta\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"interfaze/interfaze-beta\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5": {
"id": "xiaomi/mimo-v2.5",
"name": "MiMo M2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1050000,
"output": 131100
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/xiaomi/mimo-v2.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi/mimo-v2.5-pro": {
"id": "xiaomi/mimo-v2.5-pro",
"name": "MiMo V2.5 Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1050000,
"output": 131000
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.0036
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/xiaomi/mimo-v2.5-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi/mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.1-lightning": {
"id": "minimax/minimax-m2.1-lightning",
"name": "MiniMax M2.1 Lightning",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-12-23",
"last_updated": "2025-10-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 2.4,
"cache_read": 0.03,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/minimax/minimax-m2.1-lightning\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.1-lightning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-h3": {
"id": "minimax/minimax-h3",
"name": "MiniMax H3",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "minimax",
"attachment": true,
"reasoning": false,
"tool_call": false,
"release_date": "2026-07-30",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/minimax/minimax-h3\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-h3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.1": {
"id": "minimax/minimax-m2.1",
"name": "MiniMax M2.1",
"description": "Earlier MiniMax agent model for practical coding and productivity tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": true,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/minimax/minimax-m2.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2": {
"id": "minimax/minimax-m2",
"name": "MiniMax M2",
"description": "Efficient open MiniMax model built for coding agents and tool-heavy workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-10-27",
"last_updated": "2025-10-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 205000,
"output": 205000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/minimax/minimax-m2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7-highspeed": {
"id": "minimax/minimax-m2.7-highspeed",
"name": "MiniMax M2.7 High Speed",
"description": "Low-latency M2.7 variant for interactive coding plans and agent loops",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131100
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/minimax/minimax-m2.7-highspeed\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7": {
"id": "minimax/minimax-m2.7",
"name": "Minimax M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/minimax/minimax-m2.7\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5": {
"id": "minimax/minimax-m2.5",
"name": "MiniMax M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/minimax/minimax-m2.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m3": {
"id": "minimax/minimax-m3",
"name": "MiniMax M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512000,
"output": 512000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/minimax/minimax-m3\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5-highspeed": {
"id": "minimax/minimax-m2.5-highspeed",
"name": "MiniMax M2.5 High Speed",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-13",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131000
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.03,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/minimax/minimax-m2.5-highspeed\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-h3-max": {
"id": "minimax/minimax-h3-max",
"name": "MiniMax H3 Max",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "minimax",
"attachment": true,
"reasoning": false,
"tool_call": false,
"release_date": "2026-08-27",
"last_updated": "2026-08-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/minimax/minimax-h3-max\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-h3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.7-max": {
"id": "alibaba/qwen3.7-max",
"name": "Qwen 3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1,
"max": 262144
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"output": 64000
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.5,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.7-max\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-vl-thinking": {
"id": "alibaba/qwen3-vl-thinking",
"name": "Qwen3 VL Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1,
"max": 81920
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-09-23",
"last_updated": "2025-09-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-vl-thinking\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-vl-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen-3-235b": {
"id": "alibaba/qwen-3-235b",
"name": "Qwen3 235B A22B Instruct 2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04-28",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.22,
"output": 0.88
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen-3-235b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen-3-235b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-coder-plus": {
"id": "alibaba/qwen3-coder-plus",
"name": "Qwen3 Coder Plus",
"description": "Hosted Qwen coder for software agents, repo edits, and long-context code",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-coder-plus\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-next-80b-a3b-thinking": {
"id": "alibaba/qwen3-next-80b-a3b-thinking",
"name": "Qwen3 Next 80B A3B Thinking",
"description": "Efficient Qwen thinking model for local reasoning, math, and coding agents",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-next-80b-a3b-thinking\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-next-80b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.8-flash-next": {
"id": "alibaba/qwen3.8-flash-next",
"name": "Qwen 3.8 Flash Next",
"description": "Open-weight experimental preview of the Qwen4 architecture: hybrid-attention MoE (125B total, 6B active) with vision encoder for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-27",
"last_updated": "2026-08-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.12,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.8-flash-next\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.8-flash-next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-coder-30b-a3b": {
"id": "alibaba/qwen3-coder-30b-a3b",
"name": "Qwen 3 Coder 30B A3B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-31",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 8192
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-coder-30b-a3b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-coder-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-next-80b-a3b-instruct": {
"id": "alibaba/qwen3-next-80b-a3b-instruct",
"name": "Qwen3 Next 80B A3B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/wan-v3.0-video": {
"id": "alibaba/wan-v3.0-video",
"name": "Wan v3.0 Video",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "o",
"attachment": true,
"reasoning": false,
"tool_call": false,
"release_date": "2026-08-23",
"last_updated": "2026-08-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/wan-v3.0-video\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/wan-v3.0-video\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.6-plus": {
"id": "alibaba/qwen3.6-plus",
"name": "Qwen 3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1,
"max": 131072
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.1,
"cache_write": 0.625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.6-plus\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/wan-v2.7-r2v": {
"id": "alibaba/wan-v2.7-r2v",
"name": "Wan v2.7 Reference-to-Video",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/wan-v2.7-r2v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/wan-v2.7-r2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.8-27b": {
"id": "alibaba/qwen3.8-27b",
"name": "Qwen3.8 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.1,
"cache_write": 0.625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.8-27b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/wan-v2.6-t2v": {
"id": "alibaba/wan-v2.6-t2v",
"name": "Wan v2.6 Text-to-Video",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-16",
"last_updated": "2025-12-16",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/wan-v2.6-t2v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/wan-v2.6-t2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/wan-v3.0-video-prime": {
"id": "alibaba/wan-v3.0-video-prime",
"name": "Wan v3.0 Video Prime",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "o",
"attachment": true,
"reasoning": false,
"tool_call": false,
"release_date": "2026-08-28",
"last_updated": "2026-08-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/wan-v3.0-video-prime\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/wan-v3.0-video-prime\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen-3-14b": {
"id": "alibaba/qwen-3-14b",
"name": "Qwen3-14B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04-28",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 40960,
"output": 16384
},
"cost": {
"input": 0.12,
"output": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen-3-14b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen-3-14b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-vl-instruct": {
"id": "alibaba/qwen3-vl-instruct",
"name": "Qwen3 VL Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 129024
},
"cost": {
"input": 0.4,
"output": 1.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-vl-instruct\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-vl-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-235b-a22b-thinking": {
"id": "alibaba/qwen3-235b-a22b-thinking",
"name": "Qwen3 235B A22B Thinking 2507",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1,
"max": 81920
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-04",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-235b-a22b-thinking\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-235b-a22b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.5-flash": {
"id": "alibaba/qwen3.5-flash",
"name": "Qwen 3.5 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1,
"max": 81920
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.001,
"cache_write": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.5-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-coder": {
"id": "alibaba/qwen3-coder",
"name": "Qwen3 Coder 480B A35B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-22",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 7.5,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-coder\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-coder\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-coder-next": {
"id": "alibaba/qwen3-coder-next",
"name": "Qwen3 Coder Next",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2026-02-03",
"last_updated": "2026-02-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.5,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-coder-next\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-coder-next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-embedding-4b": {
"id": "alibaba/qwen3-embedding-4b",
"name": "Qwen3 Embedding 4B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-06-05",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-embedding-4b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-embedding-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-max-preview": {
"id": "alibaba/qwen3-max-preview",
"name": "Qwen3 Max Preview",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-05",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 1.2,
"output": 6,
"cache_read": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-max-preview\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-embedding-8b": {
"id": "alibaba/qwen3-embedding-8b",
"name": "Qwen3 Embedding 8B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-06-05",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-embedding-8b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-embedding-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.6-27b": {
"id": "alibaba/qwen3.6-27b",
"name": "Qwen 3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1,
"max": 131072
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.6-27b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/wan-v2.6-r2v": {
"id": "alibaba/wan-v2.6-r2v",
"name": "Wan v2.6 Reference-to-Video",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-16",
"last_updated": "2025-12-16",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/wan-v2.6-r2v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/wan-v2.6-r2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.7-flash": {
"id": "alibaba/qwen3.7-flash",
"name": "Qwen 3.7 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"output": 64000
},
"cost": {
"input": 0.03,
"output": 0.13,
"cache_read": 0.006,
"cache_write": 0.038
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.7-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-max-thinking": {
"id": "alibaba/qwen3-max-thinking",
"name": "Qwen 3 Max Thinking",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1,
"max": 81920
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-23",
"last_updated": "2025-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 1.2,
"output": 6,
"cache_read": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-max-thinking\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-max-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.8-max-0902": {
"id": "alibaba/qwen3.8-max-0902",
"name": "Qwen3.8 Max 0902",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens",
"min": 0,
"max": 262144
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"output": 128000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.8-max-0902\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.8-max-0902\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-max": {
"id": "alibaba/qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 1.2,
"output": 6,
"cache_read": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-max\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/wan-v2.6-i2v-flash": {
"id": "alibaba/wan-v2.6-i2v-flash",
"name": "Wan v2.6 Image-to-Video Flash",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-16",
"last_updated": "2025-12-16",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/wan-v2.6-i2v-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/wan-v2.6-i2v-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.8-2.4t-a95b": {
"id": "alibaba/qwen3.8-2.4t-a95b",
"name": "Qwen3.8 2.4T A95B",
"description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 128000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.8-2.4t-a95b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.8-2.4t-a95b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/wan-v2.6-r2v-flash": {
"id": "alibaba/wan-v2.6-r2v-flash",
"name": "Wan v2.6 Reference-to-Video Flash",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-16",
"last_updated": "2025-12-16",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/wan-v2.6-r2v-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/wan-v2.6-r2v-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/wan-v2.5-t2v-preview": {
"id": "alibaba/wan-v2.5-t2v-preview",
"name": "Wan v2.5 Text-to-Video Preview",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-09-24",
"last_updated": "2025-09-24",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/wan-v2.5-t2v-preview\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/wan-v2.5-t2v-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen-3-30b": {
"id": "alibaba/qwen-3-30b",
"name": "Qwen3-30B-A3B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04-28",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 40960,
"output": 16384
},
"cost": {
"input": 0.12,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen-3-30b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen-3-30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.8-flash": {
"id": "alibaba/qwen3.8-flash",
"name": "Qwen 3.8 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 991000,
"output": 128000
},
"cost": {
"input": 0.16,
"output": 0.47,
"cache_read": 0.016,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.8-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/wan-v2.6-i2v": {
"id": "alibaba/wan-v2.6-i2v",
"name": "Wan v2.6 Image-to-Video",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-16",
"last_updated": "2025-12-16",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/wan-v2.6-i2v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/wan-v2.6-i2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.8-max": {
"id": "alibaba/qwen3.8-max",
"name": "Qwen 3.8 Max",
"description": "Preview Qwen flagship for million-token multimodal reasoning and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens",
"min": 0,
"max": 262144
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-19",
"last_updated": "2026-07-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.8-max\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen-3.6-max-preview": {
"id": "alibaba/qwen-3.6-max-preview",
"name": "Qwen 3.6 Max Preview",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1,
"max": 131072
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-04-20",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 240000,
"output": 64000
},
"cost": {
"input": 1.3,
"output": 7.8,
"cache_read": 0.26,
"cache_write": 1.625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen-3.6-max-preview\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen-3.6-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-embedding-0.6b": {
"id": "alibaba/qwen3-embedding-0.6b",
"name": "Qwen3 Embedding 0.6B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-11-14",
"last_updated": "2025-11-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-embedding-0.6b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-embedding-0.6b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-vl-235b-a22b-instruct": {
"id": "alibaba/qwen3-vl-235b-a22b-instruct",
"name": "Qwen3 VL 235B A22B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 129024
},
"cost": {
"input": 0.4,
"output": 1.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3-vl-235b-a22b-instruct\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3-vl-235b-a22b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.7-plus": {
"id": "alibaba/qwen3.7-plus",
"name": "Qwen 3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1,
"max": 262144
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.08,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.7-plus\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen-3-32b": {
"id": "alibaba/qwen-3-32b",
"name": "Qwen 3.32B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1,
"max": 38912
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04-28",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.16,
"output": 0.64
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen-3-32b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen-3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/wan-v2.7-t2v": {
"id": "alibaba/wan-v2.7-t2v",
"name": "Wan v2.7 Text-to-Video",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/wan-v2.7-t2v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/wan-v2.7-t2v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.5-plus": {
"id": "alibaba/qwen3.5-plus",
"name": "Qwen 3.5 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1,
"max": 81920
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.4,
"output": 2.4,
"cache_read": 0.04,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/alibaba/qwen3.5-plus\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"alibaba/qwen3.5-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3.5-lightning": {
"id": "nvidia/nemotron-3.5-lightning",
"name": "Nemotron 3.5 Lightning 30B",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-11",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.2,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/nvidia/nemotron-3.5-lightning\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3.5-lightning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-nano-9b-v2": {
"id": "nvidia/nemotron-nano-9b-v2",
"name": "Nvidia Nemotron Nano 9B V2",
"description": "Compact Nemotron model for efficient reasoning and deployable AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-18",
"last_updated": "2025-08-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.06,
"output": 0.23
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/nvidia/nemotron-nano-9b-v2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-nano-9b-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-nano-12b-v2-vl": {
"id": "nvidia/nemotron-nano-12b-v2-vl",
"name": "Nvidia Nemotron Nano 12B V2 VL",
"description": "Nemotron multimodal model for visual reasoning and agentic AI workflows",
"family": "nemotron",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-10-28",
"last_updated": "2025-10-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.2,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/nvidia/nemotron-nano-12b-v2-vl\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-nano-12b-v2-vl\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-super-120b-a12b": {
"id": "nvidia/nemotron-3-super-120b-a12b",
"name": "NVIDIA Nemotron 3 Super 120B A12B",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": false,
"temperature": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0.15,
"output": 0.65
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/nvidia/nemotron-3-super-120b-a12b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-super-120b-a12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-ultra-550b-a55b": {
"id": "nvidia/nemotron-3-ultra-550b-a55b",
"name": "Nemotron 3 Ultra",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 65000
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/nvidia/nemotron-3-ultra-550b-a55b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-ultra-550b-a55b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-nano-30b-a3b": {
"id": "nvidia/nemotron-3-nano-30b-a3b",
"name": "Nemotron 3 Nano 30B A3B",
"description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": false,
"temperature": true,
"release_date": "2025-12-15",
"last_updated": "2025-12-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.05,
"output": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/nvidia/nemotron-3-nano-30b-a3b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-nano-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-voice-think-fast-2.0": {
"id": "spacexai/grok-voice-think-fast-2.0",
"name": "Grok Voice Think Fast 2.0",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-07-29",
"last_updated": "2026-07-29",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-voice-think-fast-2.0\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-voice-think-fast-2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-4.20-reasoning": {
"id": "spacexai/grok-4.20-reasoning",
"name": "Grok 4.20 Reasoning",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-03-10",
"last_updated": "2026-03-10",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-4.20-reasoning\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-4.20-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-4.20-multi-agent": {
"id": "spacexai/grok-4.20-multi-agent",
"name": "Grok 4.20 Multi-Agent",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-03-10",
"last_updated": "2026-03-10",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-4.20-multi-agent\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-4.20-multi-agent\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-4.3": {
"id": "spacexai/grok-4.3",
"name": "Grok 4.3",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-4.3\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-4.20-reasoning-beta": {
"id": "spacexai/grok-4.20-reasoning-beta",
"name": "Grok 4.20 Beta Reasoning",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-4.20-reasoning-beta\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-4.20-reasoning-beta\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-imagine-image": {
"id": "spacexai/grok-imagine-image",
"name": "Grok Imagine Image",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-01-28",
"last_updated": "2026-01-28",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-imagine-image\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-imagine-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-tts": {
"id": "spacexai/grok-tts",
"name": "Grok TTS",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-tts\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-tts\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-4.20-non-reasoning": {
"id": "spacexai/grok-4.20-non-reasoning",
"name": "Grok 4.20 Non-Reasoning",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"release_date": "2026-03-10",
"last_updated": "2026-03-10",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-4.20-non-reasoning\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-4.20-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-imagine-video": {
"id": "spacexai/grok-imagine-video",
"name": "Grok Imagine",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-01-28",
"last_updated": "2026-01-28",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-imagine-video\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-imagine-video\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-4.5": {
"id": "spacexai/grok-4.5",
"name": "Grok 4.5",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-4.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-build-0.1": {
"id": "spacexai/grok-build-0.1",
"name": "Grok Build 0.1",
"description": "Grok coding model for agentic engineering, edits, and codebase workflows",
"family": "grok-build",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 1,
"output": 2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-build-0.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-build-0.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-4.20-multi-agent-beta": {
"id": "spacexai/grok-4.20-multi-agent-beta",
"name": "Grok 4.20 Multi Agent Beta",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-4.20-multi-agent-beta\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-4.20-multi-agent-beta\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-4.20-non-reasoning-beta": {
"id": "spacexai/grok-4.20-non-reasoning-beta",
"name": "Grok 4.20 Beta Non-Reasoning",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-4.20-non-reasoning-beta\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-4.20-non-reasoning-beta\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-4.1-fast-non-reasoning": {
"id": "spacexai/grok-4.1-fast-non-reasoning",
"name": "Grok 4.1 Fast Non-Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-4.1-fast-non-reasoning\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-4.1-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-4.1-fast-reasoning": {
"id": "spacexai/grok-4.1-fast-reasoning",
"name": "Grok 4.1 Fast Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-4.1-fast-reasoning\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-4.1-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-imagine-video-1.5": {
"id": "spacexai/grok-imagine-video-1.5",
"name": "Grok Imagine Video 1.5",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-05-30",
"last_updated": "2026-05-30",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-imagine-video-1.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-imagine-video-1.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-imagine-image-2.0": {
"id": "spacexai/grok-imagine-image-2.0",
"name": "Grok Imagine Image 2.0",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-08-07",
"last_updated": "2026-08-07",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-imagine-image-2.0\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-imagine-image-2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-4.6": {
"id": "spacexai/grok-4.6",
"name": "Grok 4.6",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-4.6\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-stt": {
"id": "spacexai/grok-stt",
"name": "Grok STT",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-stt\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-stt\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"spacexai/grok-voice-think-fast-1.0": {
"id": "spacexai/grok-voice-think-fast-1.0",
"name": "Grok Voice Think Fast 1.0",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "grok",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/spacexai/grok-voice-think-fast-1.0\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"spacexai/grok-voice-think-fast-1.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"prodia/flux-fast-schnell": {
"id": "prodia/flux-fast-schnell",
"name": "Flux Schnell",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-08-02",
"last_updated": "2026-06-08",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 512,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/prodia/flux-fast-schnell\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"prodia/flux-fast-schnell\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.8": {
"id": "anthropic/claude-opus-4.8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-opus-4.8\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-5-fast": {
"id": "anthropic/claude-opus-5-fast",
"name": "Claude Opus 5 (Fast)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-opus-5-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-5-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.7": {
"id": "anthropic/claude-opus-4.7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-opus-4.7\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-5": {
"id": "anthropic/claude-opus-5",
"name": "Claude Opus 5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-opus-5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.6": {
"id": "anthropic/claude-sonnet-4.6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75,
"tiers": [
{
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 6,
"output": 22.5,
"cache_read": 0.6,
"cache_write": 7.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-sonnet-4.6\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-3-haiku": {
"id": "anthropic/claude-3-haiku",
"name": "Claude Haiku 3",
"description": "Legacy model retained for compatibility with older integrations",
"family": "claude-haiku",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-08-31",
"release_date": "2024-03-13",
"last_updated": "2024-03-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 4096
},
"cost": {
"input": 0.25,
"output": 1.25,
"cache_read": 0.03,
"cache_write": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-3-haiku\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-3-haiku\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-4.5": {
"id": "anthropic/claude-haiku-4.5",
"name": "Claude Haiku 4.5",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-haiku-4.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.8-fast": {
"id": "anthropic/claude-opus-4.8-fast",
"name": "Claude Opus 4.8 (Fast)",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-opus-4.8-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.8-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.6": {
"id": "anthropic/claude-opus-4.6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-opus-4.6\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5": {
"id": "anthropic/claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-fable-5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4": {
"id": "anthropic/claude-opus-4",
"name": "Claude Opus 4",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-opus-4\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.5": {
"id": "anthropic/claude-opus-4.5",
"name": "Claude Opus 4.5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-opus-4.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4": {
"id": "anthropic/claude-sonnet-4",
"name": "Claude Sonnet 4",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 8192
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-sonnet-4\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-5": {
"id": "anthropic/claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-sonnet-5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5.1": {
"id": "anthropic/claude-fable-5.1",
"name": "Claude Fable 5.1",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-fable-5.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.5": {
"id": "anthropic/claude-sonnet-4.5",
"name": "Claude Sonnet 4.5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/anthropic/claude-sonnet-4.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-26b-a4b-it": {
"id": "google/gemma-4-26b-a4b-it",
"name": "Gemma 4 26B A4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-flash-lite-image": {
"id": "google/gemini-3.1-flash-lite-image",
"name": "Gemini 3.1 Flash Lite Image (Nano Banana 2 Lite)",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 4096
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.1-flash-lite-image\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-flash-lite-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-transcribe-live": {
"id": "google/gemini-3.5-transcribe-live",
"name": "Gemini 3.5 Transcribe Live",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "gemini",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.5-transcribe-live\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-transcribe-live\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash-image": {
"id": "google/gemini-2.5-flash-image",
"name": "Nano Banana (Gemini 2.5 Flash Image)",
"description": "Nano Banana image model for fast generation, edits, and character-consistent assets",
"family": "gemini-flash",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-08-26",
"last_updated": "2025-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-2.5-flash-image\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3-pro-image": {
"id": "google/gemini-3-pro-image",
"name": "Nano Banana Pro",
"description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits",
"family": "gemini-pro",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 32768
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3-pro-image\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3-pro-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-transcribe": {
"id": "google/gemini-3.5-transcribe",
"name": "Gemini 3.5 Transcribe",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "gemini",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 2,
"output": 12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.5-transcribe\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-transcribe\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-pro-preview": {
"id": "google/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash-lite": {
"id": "google/gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash Lite",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 512,
"max": 24576
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.6-flash": {
"id": "google/gemini-3.6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.6-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo-3.1-fast-generate-001": {
"id": "google/veo-3.1-fast-generate-001",
"name": "Veo 3.1 Fast Generate",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-10-15",
"last_updated": "2026-06-08",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/veo-3.1-fast-generate-001\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/veo-3.1-fast-generate-001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-flash-lite": {
"id": "google/gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65000
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/text-embedding-005": {
"id": "google/text-embedding-005",
"name": "Text Embedding 005",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-08-01",
"last_updated": "2024-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/text-embedding-005\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/text-embedding-005\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-flash": {
"id": "google/gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.5-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-embedding-001": {
"id": "google/gemini-embedding-001",
"name": "Gemini Embedding 001",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "gemini",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-05-20",
"last_updated": "2025-05-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-embedding-001\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-embedding-001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo-3.0-generate-001": {
"id": "google/veo-3.0-generate-001",
"name": "Veo 3.0",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-05-20",
"last_updated": "2026-06-08",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/veo-3.0-generate-001\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/veo-3.0-generate-001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-flash-image": {
"id": "google/gemini-3.1-flash-image",
"name": "Nano Banana 2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.1-flash-image\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-flash-image\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-flash-lite": {
"id": "google/gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65000
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo-3.1-generate-001": {
"id": "google/veo-3.1-generate-001",
"name": "Veo 3.1",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-10-15",
"last_updated": "2026-06-08",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/veo-3.1-generate-001\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/veo-3.1-generate-001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31b-it": {
"id": "google/gemma-4-31b-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemma-4-31b-it\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-embedding-2": {
"id": "google/gemini-embedding-2",
"name": "Gemini Embedding 2",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "gemini",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-11",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-embedding-2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-embedding-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.8-flash": {
"id": "google/gemini-3.8-flash",
"name": "Gemini 3.8 Flash",
"description": "Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.8-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.7-flash": {
"id": "google/gemini-3.7-flash",
"name": "Gemini 3.7 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.7-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo-3.0-fast-generate-001": {
"id": "google/veo-3.0-fast-generate-001",
"name": "Veo 3.0 Fast Generate",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-07-31",
"last_updated": "2026-06-08",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/veo-3.0-fast-generate-001\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/veo-3.0-fast-generate-001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/text-multilingual-embedding-002": {
"id": "google/text-multilingual-embedding-002",
"name": "Text Multilingual Embedding 002",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-03-01",
"last_updated": "2024-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/text-multilingual-embedding-002\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/text-multilingual-embedding-002\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-flash-image-preview": {
"id": "google/gemini-3.1-flash-image-preview",
"name": "Gemini 3.1 Flash Image Preview (Nano Banana 2)",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-26",
"last_updated": "2026-02-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3.1-flash-image-preview\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-flash-image-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3-flash": {
"id": "google/gemini-3-flash",
"name": "Gemini 3 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65000
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-3-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-omni-flash-preview": {
"id": "google/gemini-omni-flash-preview",
"name": "Gemini Omni Flash Preview",
"description": "Omni-modal model for text, vision, audio, and multimodal agent tasks",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 57920
},
"cost": {
"input": 1.5,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-omni-flash-preview\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-omni-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo-3.1-lite-generate-001": {
"id": "google/veo-3.1-lite-generate-001",
"name": "Veo 3.1 Lite Generate",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/veo-3.1-lite-generate-001\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/veo-3.1-lite-generate-001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash": {
"id": "google/gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"input_audio": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-2.5-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-pro": {
"id": "google/gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/google/gemini-2.5-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bfl/flux-kontext-max": {
"id": "bfl/flux-kontext-max",
"name": "FLUX.1 Kontext Max",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-05-29",
"last_updated": "2025-06",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 512,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bfl/flux-kontext-max\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bfl/flux-kontext-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bfl/flux-2-pro": {
"id": "bfl/flux-2-pro",
"name": "FLUX.2 [pro]",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-11-25",
"last_updated": "2026-06-08",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 67300,
"output": 67300
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bfl/flux-2-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bfl/flux-2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bfl/flux-2-max": {
"id": "bfl/flux-2-max",
"name": "FLUX.2 [max]",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-16",
"last_updated": "2026-06-08",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 67300,
"output": 67300
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bfl/flux-2-max\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bfl/flux-2-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bfl/flux-pro-1.1-ultra": {
"id": "bfl/flux-pro-1.1-ultra",
"name": "FLUX1.1 [pro] Ultra",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-11-01",
"last_updated": "2024-11",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 512,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bfl/flux-pro-1.1-ultra\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bfl/flux-pro-1.1-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bfl/flux-kontext-pro": {
"id": "bfl/flux-kontext-pro",
"name": "FLUX.1 Kontext Pro",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-05-29",
"last_updated": "2025-06",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 512,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bfl/flux-kontext-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bfl/flux-kontext-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bfl/flux-3-video": {
"id": "bfl/flux-3-video",
"name": "Flux 3",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-08-04",
"last_updated": "2026-08-04",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bfl/flux-3-video\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bfl/flux-3-video\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bfl/flux-pro-1.0-fill": {
"id": "bfl/flux-pro-1.0-fill",
"name": "FLUX.1 Fill [pro]",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-10-01",
"last_updated": "2024-10",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 512,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bfl/flux-pro-1.0-fill\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bfl/flux-pro-1.0-fill\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bfl/flux-2-flex": {
"id": "bfl/flux-2-flex",
"name": "FLUX.2 [flex]",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-11-25",
"last_updated": "2026-06-08",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bfl/flux-2-flex\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bfl/flux-2-flex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bfl/flux-2-klein-9b": {
"id": "bfl/flux-2-klein-9b",
"name": "FLUX.2 [klein] 9B",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-01-15",
"last_updated": "2026-06-08",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bfl/flux-2-klein-9b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bfl/flux-2-klein-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bfl/flux-2-klein-4b": {
"id": "bfl/flux-2-klein-4b",
"name": "FLUX.2 [klein] 4B",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-01-15",
"last_updated": "2026-06-08",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bfl/flux-2-klein-4b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bfl/flux-2-klein-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bfl/flux-pro-1.1": {
"id": "bfl/flux-pro-1.1",
"name": "FLUX1.1 [pro]",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "flux",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-10-02",
"last_updated": "2024-10",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 512,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bfl/flux-pro-1.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bfl/flux-pro-1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/inkling-small": {
"id": "thinkingmachines/inkling-small",
"name": "Inkling Small",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-30",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 0.5,
"output": 1.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/thinkingmachines/inkling-small\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/inkling-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/inkling": {
"id": "thinkingmachines/inkling",
"name": "Inkling",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 1,
"output": 4.05,
"cache_read": 0.17
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/thinkingmachines/inkling\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.3": {
"id": "meta/muse-spark-1.3",
"name": "Muse Spark 1.3",
"description": "Open Llama multimodal model for image understanding and text reasoning",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/muse-spark-1.3\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.2": {
"id": "meta/muse-spark-1.2",
"name": "Muse Spark 1.2",
"description": "Open Llama multimodal model for image understanding and text reasoning",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/muse-spark-1.2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.2-contributor": {
"id": "meta/muse-spark-1.2-contributor",
"name": "Muse Spark 1.2 Contributor",
"description": "Open Llama multimodal model for image understanding and text reasoning",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.002
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/muse-spark-1.2-contributor\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.2-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-3.1-8b": {
"id": "meta/llama-3.1-8b",
"name": "Llama 3.1 8B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.22,
"output": 0.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/llama-3.1-8b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-3.1-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.3-contributor": {
"id": "meta/muse-spark-1.3-contributor",
"name": "Muse Spark 1.3 Contributor",
"description": "Open Llama multimodal model for image understanding and text reasoning",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.002
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/muse-spark-1.3-contributor\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.3-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.1": {
"id": "meta/muse-spark-1.1",
"name": "Muse Spark 1.1",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "muse",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-08",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/muse-spark-1.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-glimmer-30b": {
"id": "meta/muse-glimmer-30b",
"name": "Muse Glimmer 30B",
"description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-01-04",
"release_date": "2026-08-10",
"last_updated": "2026-08-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.35,
"output": 1.5,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/muse-glimmer-30b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-glimmer-30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-3.1-70b": {
"id": "meta/llama-3.1-70b",
"name": "Llama 3.1 70B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.72,
"output": 0.72
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/llama-3.1-70b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-3.1-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-image-1.0": {
"id": "meta/muse-image-1.0",
"name": "Muse Image 1.0",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "muse",
"attachment": true,
"reasoning": false,
"tool_call": false,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/muse-image-1.0\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-image-1.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-3.3-70b": {
"id": "meta/llama-3.3-70b",
"name": "Llama-3.3-70B-Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/llama-3.3-70b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-3.3-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-4-scout": {
"id": "meta/llama-4-scout",
"name": "Llama-4-Scout-17B-16E-Instruct-FP8",
"description": "Open multimodal Llama model for long-context analysis and efficient agents",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/llama-4-scout\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-4-scout\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/llama-4-maverick": {
"id": "meta/llama-4-maverick",
"name": "Llama-4-Maverick-17B-128E-Instruct-FP8",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/meta/llama-4-maverick\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"meta/llama-4-maverick\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"quiverai/arrow-1.1": {
"id": "quiverai/arrow-1.1",
"name": "Arrow 1.1",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/quiverai/arrow-1.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"quiverai/arrow-1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedance-2.0-mini": {
"id": "bytedance/seedance-2.0-mini",
"name": "Seedance 2.0 Mini",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "seed",
"attachment": true,
"reasoning": false,
"tool_call": false,
"release_date": "2026-06-22",
"last_updated": "2026-06-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seedance-2.0-mini\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedance-2.0-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedance-v1.0-pro-fast": {
"id": "bytedance/seedance-v1.0-pro-fast",
"name": "Seedance v1.0 Pro Fast",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-10-24",
"last_updated": "2025-10-31",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seedance-v1.0-pro-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedance-v1.0-pro-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedance-v1.0-pro": {
"id": "bytedance/seedance-v1.0-pro",
"name": "Seedance v1.0 Pro",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-06-11",
"last_updated": "2025-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seedance-v1.0-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedance-v1.0-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedance-v1.5-pro": {
"id": "bytedance/seedance-v1.5-pro",
"name": "Seedance v1.5 Pro",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-16",
"last_updated": "2025-12-16",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seedance-v1.5-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedance-v1.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedream-4.5": {
"id": "bytedance/seedream-4.5",
"name": "Seedream 4.5",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-03",
"last_updated": "2025-11-28",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seedream-4.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedream-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seed-1.8": {
"id": "bytedance/seed-1.8",
"name": "Seed 1.8",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-09-01",
"last_updated": "2025-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seed-1.8\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seed-1.8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seed-1.6": {
"id": "bytedance/seed-1.6",
"name": "Seed 1.6",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-09-01",
"last_updated": "2025-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seed-1.6\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seed-1.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedance-2.0-fast": {
"id": "bytedance/seedance-2.0-fast",
"name": "Seedance 2.0 Fast",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "seed",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-04-14",
"last_updated": "2026-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seedance-2.0-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedance-2.0-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedream-4.0": {
"id": "bytedance/seedream-4.0",
"name": "Seedream 4.0",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-09-09",
"last_updated": "2025-08-28",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seedream-4.0\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedream-4.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedream-5.0-lite": {
"id": "bytedance/seedream-5.0-lite",
"name": "Seedream 5.0 Lite",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-02-13",
"last_updated": "2026-01-28",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seedream-5.0-lite\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedream-5.0-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedance-2.5": {
"id": "bytedance/seedance-2.5",
"name": "Seedance 2.5",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-08-07",
"last_updated": "2026-08-07",
"modalities": {
"input": [
"text"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seedance-2.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedance-2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedance-2.0": {
"id": "bytedance/seedance-2.0",
"name": "Seedance 2.0",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "seed",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-04-14",
"last_updated": "2026-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seedance-2.0\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedance-2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedream-5.0-pro": {
"id": "bytedance/seedream-5.0-pro",
"name": "Seedream 5.0 Pro",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "seed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-07-11",
"last_updated": "2026-07-11",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/bytedance/seedream-5.0-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedream-5.0-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inception/mercury-coder-small": {
"id": "inception/mercury-coder-small",
"name": "Mercury Coder Small Beta",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"family": "mercury",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-02-26",
"last_updated": "2025-02-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 16384
},
"cost": {
"input": 0.25,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/inception/mercury-coder-small\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"inception/mercury-coder-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inception/mercury-2.5": {
"id": "inception/mercury-2.5",
"name": "Mercury 2.5",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"family": "mercury",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"release_date": "2026-09-08",
"last_updated": "2026-09-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 260000,
"output": 65536
},
"cost": {
"input": 0.04,
"output": 0.15,
"cache_read": 0.004
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/inception/mercury-2.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"inception/mercury-2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inception/mercury-2": {
"id": "inception/mercury-2",
"name": "Mercury 2",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"family": "mercury",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-24",
"last_updated": "2026-03-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 0.75,
"cache_read": 0.024999999999999998
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/inception/mercury-2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"inception/mercury-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fish-audio/transcribe-1": {
"id": "fish-audio/transcribe-1",
"name": "Transcribe-1",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-03-01",
"last_updated": "2026-03-01",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/fish-audio/transcribe-1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fish-audio/transcribe-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fish-audio/s2.1-pro-free": {
"id": "fish-audio/s2.1-pro-free",
"name": "S2.1 Pro (Free)",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-07-28",
"last_updated": "2026-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/fish-audio/s2.1-pro-free\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fish-audio/s2.1-pro-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fish-audio/s2.1-pro": {
"id": "fish-audio/s2.1-pro",
"name": "S2.1 Pro",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-07-28",
"last_updated": "2026-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/fish-audio/s2.1-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fish-audio/s2.1-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fish-audio/s1-free": {
"id": "fish-audio/s1-free",
"name": "S1 (Free)",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2025-10-20",
"last_updated": "2025-10-20",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/fish-audio/s1-free\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fish-audio/s1-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fish-audio/s2-pro": {
"id": "fish-audio/s2-pro",
"name": "S2 Pro",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/fish-audio/s2-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fish-audio/s2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fish-audio/s2-pro-free": {
"id": "fish-audio/s2-pro-free",
"name": "S2 Pro (Free)",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/fish-audio/s2-pro-free\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fish-audio/s2-pro-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fish-audio/transcribe-1-free": {
"id": "fish-audio/transcribe-1-free",
"name": "Transcribe-1 (Free)",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-03-01",
"last_updated": "2026-03-01",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/fish-audio/transcribe-1-free\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fish-audio/transcribe-1-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fish-audio/s1": {
"id": "fish-audio/s1",
"name": "S1",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2025-10-20",
"last_updated": "2025-10-20",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/fish-audio/s1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fish-audio/s1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sakana/fugu-ultra": {
"id": "sakana/fugu-ultra",
"name": "Fugu Ultra",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/sakana/fugu-ultra\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"sakana/fugu-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sakana/namazu": {
"id": "sakana/namazu",
"name": "Sakana Namazu",
"description": "Multi-agent model for routing expert agents across complex analytical tasks",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/sakana/namazu\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"sakana/namazu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-vision-exp": {
"id": "deepseek/deepseek-v4-flash-vision-exp",
"name": "DeepSeek V4 Flash Vision Exp",
"description": "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-21",
"last_updated": "2026-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.22,
"output": 0.66,
"cache_read": 0.007
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/deepseek/deepseek-v4-flash-vision-exp\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-vision-exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro-0813": {
"id": "deepseek/deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.66,
"output": 1.98,
"cache_read": 0.066
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/deepseek/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-0731": {
"id": "deepseek/deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.076,
"output": 0.153,
"cache_read": 0.014
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/deepseek/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.2-thinking": {
"id": "deepseek/deepseek-v3.2-thinking",
"name": "DeepSeek V3.2 Thinking",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8000
},
"cost": {
"input": 0.62,
"output": 1.85
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/deepseek/deepseek-v3.2-thinking\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.2-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash": {
"id": "deepseek/deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.13,
"output": 0.26,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/deepseek/deepseek-v4-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4.1-flash": {
"id": "deepseek/deepseek-v4.1-flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393216
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.006
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/deepseek/deepseek-v4.1-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4.1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.2": {
"id": "deepseek/deepseek-v3.2",
"name": "DeepSeek V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8000
},
"cost": {
"input": 0.62,
"output": 1.85
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/deepseek/deepseek-v3.2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.1-terminus": {
"id": "deepseek/deepseek-v3.1-terminus",
"name": "DeepSeek V3.1 Terminus",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-09-22",
"last_updated": "2025-09-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0.27,
"output": 1,
"cache_read": 0.135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/deepseek/deepseek-v3.1-terminus\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.1-terminus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro": {
"id": "deepseek/deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.66,
"output": 1.98,
"cache_read": 0.022
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/deepseek/deepseek-v4-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.1": {
"id": "deepseek/deepseek-v3.1",
"name": "DeepSeek-V3.1",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 0.95,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/deepseek/deepseek-v3.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-r1": {
"id": "deepseek/deepseek-r1",
"name": "DeepSeek-R1",
"description": "Classic open reasoning model for transparent math, coding, and deliberate problem solving",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-01-20",
"last_updated": "2025-05-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 1.35,
"output": 5.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/deepseek/deepseek-r1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-r1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon/nova-2-lite": {
"id": "amazon/nova-2-lite",
"name": "Nova 2 Lite",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "nova",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-12-02",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/amazon/nova-2-lite\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"amazon/nova-2-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon/titan-embed-text-v2": {
"id": "amazon/titan-embed-text-v2",
"name": "Titan Text Embeddings V2",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "titan-embed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-04-30",
"last_updated": "2024-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/amazon/titan-embed-text-v2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"amazon/titan-embed-text-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon/nova-pro": {
"id": "amazon/nova-pro",
"name": "Nova Pro",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "nova-pro",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-12-03",
"last_updated": "2024-12-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 300000,
"output": 10000
},
"cost": {
"input": 0.8,
"output": 3.2,
"cache_read": 0.2,
"cache_write": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/amazon/nova-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"amazon/nova-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon/nova-lite": {
"id": "amazon/nova-lite",
"name": "Nova Lite",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "nova-lite",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-12-03",
"last_updated": "2024-12-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 300000,
"output": 10000
},
"cost": {
"input": 0.06,
"output": 0.24,
"cache_read": 0.015,
"cache_write": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/amazon/nova-lite\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"amazon/nova-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"amazon/nova-micro": {
"id": "amazon/nova-micro",
"name": "Nova Micro",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "nova-micro",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-12-03",
"last_updated": "2024-12-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 10000
},
"cost": {
"input": 0.035,
"output": 0.14,
"cache_read": 0.00875,
"cache_write": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/amazon/nova-micro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"amazon/nova-micro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionai/ling-3.0-flash": {
"id": "inclusionai/ling-3.0-flash",
"name": "Ling 3.0 Flash",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "ling",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-08-06",
"last_updated": "2026-08-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0.06,
"output": 0.18,
"cache_read": 0.012
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/inclusionai/ling-3.0-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"inclusionai/ling-3.0-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionai/ling-3.0-flash-fin": {
"id": "inclusionai/ling-3.0-flash-fin",
"name": "Ling 3.0 Flash Fin",
"description": "Finance-enhanced model for financial research, multi-step investment workflows, and long-horizon planning and execution",
"family": "ling",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"release_date": "2026-08-27",
"last_updated": "2026-08-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/inclusionai/ling-3.0-flash-fin\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"inclusionai/ling-3.0-flash-fin\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionai/ling-3.0-flash-sante": {
"id": "inclusionai/ling-3.0-flash-sante",
"name": "Ling 3.0 Flash Sante",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "ling",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/inclusionai/ling-3.0-flash-sante\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"inclusionai/ling-3.0-flash-sante\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionai/ling-3.0-flash-sante-free": {
"id": "inclusionai/ling-3.0-flash-sante-free",
"name": "Ling 3.0 Flash Sante (Free)",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "ling",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/inclusionai/ling-3.0-flash-sante-free\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"inclusionai/ling-3.0-flash-sante-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inclusionai/ling-3.0-flash-fin-free": {
"id": "inclusionai/ling-3.0-flash-fin-free",
"name": "Ling 3.0 Flash Fin (Free)",
"description": "Finance-enhanced model for financial research, multi-step investment workflows, and long-horizon planning and execution",
"family": "ling",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"release_date": "2026-08-27",
"last_updated": "2026-08-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/inclusionai/ling-3.0-flash-fin-free\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"inclusionai/ling-3.0-flash-fin-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-fast": {
"id": "openai/gpt-5-fast",
"name": "GPT-5 (Fast)",
"description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 20,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-codex": {
"id": "openai/gpt-5-codex",
"name": "GPT-5-Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-09-15",
"last_updated": "2025-09-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5-codex\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-pro": {
"id": "openai/gpt-5-pro",
"name": "GPT-5 pro",
"description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 128000,
"output": 272000
},
"cost": {
"input": 15,
"output": 120
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini-transcribe": {
"id": "openai/gpt-4o-mini-transcribe",
"name": "GPT-4o mini Transcribe",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "o-mini",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-03-13",
"last_updated": "2024-03-13",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 1.25,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4o-mini-transcribe\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini-transcribe\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex-mini": {
"id": "openai/gpt-5.1-codex-mini",
"name": "GPT-5.1 Codex mini",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.1-codex-mini\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5-fast": {
"id": "openai/gpt-5.5-fast",
"name": "GPT 5.5 (Fast)",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 872000,
"output": 128000
},
"cost": {
"input": 12.5,
"output": 75,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.5-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex": {
"id": "openai/gpt-5.1-codex",
"name": "GPT-5.1-Codex",
"description": "Codex GPT for repository edits, code review, and practical software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.1-codex\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-mini-fast": {
"id": "openai/gpt-5.4-mini-fast",
"name": "GPT 5.4 Mini (Fast)",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.4-mini-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-mini-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-sol": {
"id": "openai/gpt-5.6-sol",
"name": "GPT 5.6 Sol",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.6-sol\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-realtime-1.5": {
"id": "openai/gpt-realtime-1.5",
"name": "GPT-Realtime-1.5",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 4,
"output": 16,
"cache_read": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-realtime-1.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-realtime-1.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-mini-fast": {
"id": "openai/gpt-5-mini-fast",
"name": "GPT-5 mini (Fast)",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.45,
"output": 3.6,
"cache_read": 0.045
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5-mini-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-mini-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/tts-1": {
"id": "openai/tts-1",
"name": "TTS-1",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2023-11-06",
"last_updated": "2023-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/tts-1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/tts-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-codex": {
"id": "openai/gpt-5.2-codex",
"name": "GPT-5.2-Codex",
"description": "Code-specialist GPT for repository edits, reviews, and long-running software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.2-codex\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-6-astra": {
"id": "openai/gpt-6-astra",
"name": "GPT-6 Astra",
"description": "GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.",
"family": "gpt-astra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-04-30",
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5,
"tiers": [
{
"input": 20,
"output": 75,
"cache_read": 2,
"cache_write": 25,
"tier": {
"type": "context",
"size": 272001
}
}
],
"context_over_200k": {
"input": 20,
"output": 75,
"cache_read": 2,
"cache_write": 25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-6-astra\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-6-astra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-6-astra-fast": {
"id": "openai/gpt-6-astra-fast",
"name": "GPT-6 Astra (Fast)",
"description": "Fast variant of GPT-6 Astra for low-latency assistance and high-volume workloads.",
"family": "gpt-astra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 20,
"output": 100,
"cache_read": 2,
"cache_write": 25,
"tiers": [
{
"input": 40,
"output": 150,
"cache_read": 4,
"cache_write": 25,
"tier": {
"type": "context",
"size": 272001
}
}
],
"context_over_200k": {
"input": 40,
"output": 150,
"cache_read": 4,
"cache_write": 25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-6-astra-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-6-astra-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-pro": {
"id": "openai/gpt-5.2-pro",
"name": "GPT 5.2 ",
"description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 21,
"output": 168
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.2-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-realtime-2": {
"id": "openai/gpt-realtime-2",
"name": "gpt-realtime-2",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 4,
"output": 24,
"cache_read": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-realtime-2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-realtime-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4": {
"id": "openai/gpt-5.4",
"name": "GPT 5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.4\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"input": 122880,
"output": 8192
},
"cost": {
"input": 0.05,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-oss-20b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-transcribe": {
"id": "openai/gpt-4o-transcribe",
"name": "GPT-4o Transcribe",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-03-13",
"last_updated": "2024-03-13",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 2.5,
"output": 10
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4o-transcribe\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-transcribe\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o4-mini-fast": {
"id": "openai/o4-mini-fast",
"name": "o4-mini (Fast)",
"description": "Fast o-series model for compact reasoning, coding, and tool use",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 100000,
"output": 100000
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/o4-mini-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/o4-mini-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-safeguard-20b": {
"id": "openai/gpt-oss-safeguard-20b",
"name": "gpt-oss-safeguard-20b",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-10-29",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 112000,
"output": 16000
},
"cost": {
"input": 0.07,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-oss-safeguard-20b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-safeguard-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-codex-max": {
"id": "openai/gpt-5.1-codex-max",
"name": "GPT 5.1 Codex Max",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.1-codex-max\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-codex-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-safeguard-120b": {
"id": "openai/gpt-oss-safeguard-120b",
"name": "GPT OSS Safeguard 120B",
"description": "Safety model for policy screening, moderation, and risk-aware routing workflows",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-29",
"last_updated": "2025-10-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"input": 112000,
"output": 16000
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-oss-safeguard-120b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-safeguard-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-luna": {
"id": "openai/gpt-5.6-luna",
"name": "GPT 5.6 Luna",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.6-luna\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-fast": {
"id": "openai/gpt-4.1-fast",
"name": "GPT-4.1 (Fast)",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"input": 1014808,
"output": 32768
},
"cost": {
"input": 3.5,
"output": 14,
"cache_read": 0.875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4.1-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini-fast": {
"id": "openai/gpt-4o-mini-fast",
"name": "GPT-4o mini (Fast)",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 111616,
"output": 16384
},
"cost": {
"input": 0.25,
"output": 1,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4o-mini-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-luna-fast": {
"id": "openai/gpt-5.6-luna-fast",
"name": "GPT 5.6 Luna (Fast)",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.4,
"output": 2.4,
"cache_read": 0.04,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.6-luna-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-luna-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-codex": {
"id": "openai/gpt-5.3-codex",
"name": "GPT 5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.3-codex\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-image-1.5": {
"id": "openai/gpt-image-1.5",
"name": "GPT Image 1.5",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-image",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-11-25",
"last_updated": "2025-11-25",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 5,
"output": 32,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-image-1.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-image-1.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-fast": {
"id": "openai/gpt-5.4-fast",
"name": "GPT 5.4 (Fast)",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.4-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-thinking-fast": {
"id": "openai/gpt-5.1-thinking-fast",
"name": "GPT 5.1 Thinking (Fast)",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2025-11-12",
"last_updated": "2025-11-12",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 20,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.1-thinking-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-thinking-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/text-embedding-ada-002": {
"id": "openai/text-embedding-ada-002",
"name": "text-embedding-ada-002",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2022-12-15",
"last_updated": "2022-12-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"input": 6656,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/text-embedding-ada-002\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/text-embedding-ada-002\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-image-1": {
"id": "openai/gpt-image-1",
"name": "GPT Image 1",
"description": "OpenAI image model for production generation, edits, and brand-safe visual workflows",
"family": "gpt-image",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-04-24",
"last_updated": "2025-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 5,
"output": 40,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-image-1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-image-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-nano": {
"id": "openai/gpt-5.4-nano",
"name": "GPT 5.4 Nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.4-nano\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1-thinking": {
"id": "openai/gpt-5.1-thinking",
"name": "GPT 5.1 Thinking",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-11-12",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.1-thinking\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5-pro": {
"id": "openai/gpt-5.5-pro",
"name": "GPT 5.5 Pro",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 872000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.5-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/tts-1-hd": {
"id": "openai/tts-1-hd",
"name": "TTS-1 HD",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2023-11-06",
"last_updated": "2023-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/tts-1-hd\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/tts-1-hd\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-fast": {
"id": "openai/o3-fast",
"name": "o3 (Fast)",
"description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 100000,
"output": 100000
},
"cost": {
"input": 3.5,
"output": 14,
"cache_read": 0.875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/o3-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-image-1-mini": {
"id": "openai/gpt-image-1-mini",
"name": "GPT Image 1 Mini",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-image",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-image-1-mini\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-image-1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-terra-fast": {
"id": "openai/gpt-5.6-terra-fast",
"name": "GPT 5.6 Terra (Fast)",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 4,
"output": 24,
"cache_read": 0.4,
"cache_write": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.6-terra-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-terra-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-mini": {
"id": "openai/gpt-5.4-mini",
"name": "GPT 5.4 Mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.4-mini\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-image-2.5-sunburst": {
"id": "openai/gpt-image-2.5-sunburst",
"name": "GPT Image 2.5 Sunburst",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-image",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-09-08",
"last_updated": "2026-09-08",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-image-2.5-sunburst\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-image-2.5-sunburst\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-nano-fast": {
"id": "openai/gpt-4.1-nano-fast",
"name": "GPT-4.1 nano (Fast)",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"input": 1014808,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.8,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4.1-nano-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-nano-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-realtime-mini": {
"id": "openai/gpt-realtime-mini",
"name": "GPT-Realtime mini",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-10-10",
"last_updated": "2025-10-10",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-realtime-mini\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-realtime-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-image-2": {
"id": "openai/gpt-image-2",
"name": "GPT Image 2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-image",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-image-2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-image-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-mini-fast": {
"id": "openai/gpt-4.1-mini-fast",
"name": "GPT-4.1 mini (Fast)",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"input": 1014808,
"output": 32768
},
"cost": {
"input": 0.7,
"output": 2.8,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4.1-mini-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-mini-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-realtime-whisper": {
"id": "openai/gpt-realtime-whisper",
"name": "gpt-realtime-whisper",
"description": "Streaming speech-to-text model for low-latency transcript deltas from live audio",
"family": "whisper",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-realtime-whisper\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-realtime-whisper\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-3.5-turbo": {
"id": "openai/gpt-3.5-turbo",
"name": "GPT-3.5 Turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2021-09",
"release_date": "2023-03-01",
"last_updated": "2023-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16385,
"input": 12289,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-3.5-turbo\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-3.5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-codex-fast": {
"id": "openai/gpt-5.3-codex-fast",
"name": "GPT 5.3 Codex (Fast)",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 3.5,
"output": 28,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.3-codex-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-codex-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/text-embedding-3-small": {
"id": "openai/text-embedding-3-small",
"name": "text-embedding-3-small",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-01-25",
"last_updated": "2024-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"input": 6656,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/text-embedding-3-small\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/text-embedding-3-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-oss-120b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-pro": {
"id": "openai/gpt-5.4-pro",
"name": "GPT 5.4 Pro",
"description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.4-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/text-embedding-3-large": {
"id": "openai/text-embedding-3-large",
"name": "text-embedding-3-large",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-01-25",
"last_updated": "2024-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"input": 6656,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/text-embedding-3-large\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/text-embedding-3-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-image-2.5-flare": {
"id": "openai/gpt-image-2.5-flare",
"name": "GPT Image 2.5 Flare",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-image",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-09-08",
"last_updated": "2026-09-08",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-image-2.5-flare\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-image-2.5-flare\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-terra": {
"id": "openai/gpt-5.6-terra",
"name": "GPT 5.6 Terra",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.6-terra\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2-fast": {
"id": "openai/gpt-5.2-fast",
"name": "GPT 5.2 (Fast)",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 3.5,
"output": 28,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.2-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2": {
"id": "openai/gpt-5.2",
"name": "GPT-5.2",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-sol-fast": {
"id": "openai/gpt-5.6-sol-fast",
"name": "GPT 5.6 Sol (Fast)",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 4,
"output": 20,
"cache_read": 0.4,
"cache_write": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.6-sol-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-sol-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/whisper-1": {
"id": "openai/whisper-1",
"name": "Whisper",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "whisper",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2022-09-21",
"last_updated": "2022-09-21",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/whisper-1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/whisper-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-fast": {
"id": "openai/gpt-4o-fast",
"name": "GPT-4o (Fast)",
"description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 111616,
"output": 16384
},
"cost": {
"input": 4.25,
"output": 17,
"cache_read": 2.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4o-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-realtime-2.1": {
"id": "openai/gpt-realtime-2.1",
"name": "gpt-realtime-2.1",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "gpt",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 96000,
"output": 32000
},
"cost": {
"input": 4,
"output": 24,
"cache_read": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-realtime-2.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-realtime-2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-pro": {
"id": "openai/o3-pro",
"name": "o3 Pro",
"description": "High-effort o3 tier for difficult technical reasoning and careful answers",
"family": "o-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-06-10",
"last_updated": "2025-06-10",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 100000,
"output": 100000
},
"cost": {
"input": 20,
"output": 80
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/o3-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT 5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 872000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3": {
"id": "openai/o3",
"name": "o3",
"description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/o3\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-mini": {
"id": "openai/o3-mini",
"name": "o3-mini",
"description": "Smaller o-series reasoner for economical coding, math, and planning tasks",
"family": "o-mini",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2024-12-20",
"last_updated": "2025-01-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"status": "deprecated",
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/o3-mini\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o4-mini": {
"id": "openai/o4-mini",
"name": "o4-mini",
"description": "Fast o-series model for compact reasoning, coding, and tool use",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"status": "deprecated",
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/o4-mini\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/o4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5": {
"id": "openai/gpt-5",
"name": "GPT-5",
"description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-mini": {
"id": "openai/gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5-mini\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1": {
"id": "openai/gpt-4.1",
"name": "GPT-4.1",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini": {
"id": "openai/gpt-4o-mini",
"name": "GPT-4o mini",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4o-mini\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o": {
"id": "openai/gpt-4o",
"name": "GPT-4o",
"description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4o\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o1": {
"id": "openai/o1",
"name": "o1",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2023-09",
"release_date": "2024-12-05",
"last_updated": "2024-12-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"status": "deprecated",
"cost": {
"input": 15,
"output": 60,
"cache_read": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/o1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/o1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4-turbo": {
"id": "openai/gpt-4-turbo",
"name": "GPT-4 Turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"status": "deprecated",
"cost": {
"input": 10,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4-turbo\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-mini": {
"id": "openai/gpt-4.1-mini",
"name": "GPT-4.1 mini",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4.1-mini\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-nano": {
"id": "openai/gpt-4.1-nano",
"name": "GPT-4.1 nano",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"status": "deprecated",
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-4.1-nano\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-nano": {
"id": "openai/gpt-5-nano",
"name": "GPT-5 Nano",
"description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/openai/gpt-5-nano\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.7-code-highspeed": {
"id": "moonshotai/kimi-k2.7-code-highspeed",
"name": "Kimi K2.7 Code High Speed",
"description": "Lower-latency Kimi Code variant for interactive edits and coding-agent loops",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 1.9,
"output": 8,
"cache_read": 0.38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/moonshotai/kimi-k2.7-code-highspeed\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.7-code-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.6": {
"id": "moonshotai/kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/moonshotai/kimi-k2.6\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.7-code": {
"id": "moonshotai/kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 32768
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/moonshotai/kimi-k2.7-code\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2-thinking": {
"id": "moonshotai/kimi-k2-thinking",
"name": "Kimi K2 Thinking",
"description": "Thinking Kimi model for slower research passes, planning, and hard technical questions",
"family": "kimi-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-11-06",
"last_updated": "2025-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 216144,
"output": 216144
},
"cost": {
"input": 0.47,
"output": 2,
"cache_read": 0.141
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/moonshotai/kimi-k2-thinking\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k3": {
"id": "moonshotai/kimi-k3",
"name": "Kimi K3",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/moonshotai/kimi-k3\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k3-fast": {
"id": "moonshotai/kimi-k3-fast",
"name": "Kimi K3 Fast",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 4.5,
"output": 22.5,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/moonshotai/kimi-k3-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2": {
"id": "moonshotai/kimi-k2",
"name": "Kimi K2 Instruct",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-07-11",
"last_updated": "2025-09-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.57,
"output": 2.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/moonshotai/kimi-k2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.5": {
"id": "moonshotai/kimi-k2.5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262114,
"output": 262114
},
"cost": {
"input": 0.6,
"output": 3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/moonshotai/kimi-k2.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cohere/rerank-v4-pro": {
"id": "cohere/rerank-v4-pro",
"name": "Cohere Rerank 4 Pro",
"description": "Reranking model for improving retrieval quality in search and recommendation systems",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/cohere/rerank-v4-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"cohere/rerank-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cohere/rerank-v4-fast": {
"id": "cohere/rerank-v4-fast",
"name": "Cohere Rerank 4 Fast",
"description": "Reranking model for improving retrieval quality in search and recommendation systems",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 32000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/cohere/rerank-v4-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"cohere/rerank-v4-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cohere/embed-v4.0": {
"id": "cohere/embed-v4.0",
"name": "Embed v4.0",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "cohere-embed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-04-15",
"last_updated": "2025-04-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/cohere/embed-v4.0\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"cohere/embed-v4.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cohere/command-a": {
"id": "cohere/command-a",
"name": "Command A",
"description": "Cohere command model for multilingual enterprise agents, tools, and chat",
"family": "command",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-03-13",
"last_updated": "2025-03-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 8000
},
"cost": {
"input": 2.5,
"output": 10
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/cohere/command-a\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"cohere/command-a\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cohere/rerank-v3.5": {
"id": "cohere/rerank-v3.5",
"name": "Cohere Rerank 3.5",
"description": "Reranking model for improving retrieval quality in search and recommendation systems",
"family": "o",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-12-02",
"last_updated": "2024-12-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 4096,
"output": 4096
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/cohere/rerank-v3.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"cohere/rerank-v3.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"arcee-ai/trinity-large-thinking": {
"id": "arcee-ai/trinity-large-thinking",
"name": "Trinity Large Thinking",
"description": "Reasoning-optimized 398B MoE agent model with extended thinking for long-horizon and multi-turn tool use",
"family": "trinity",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-04-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262100,
"output": 80000
},
"cost": {
"input": 0.25,
"output": 0.8999999999999999
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/arcee-ai/trinity-large-thinking\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"arcee-ai/trinity-large-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/hy-mt2-lite": {
"id": "tencent/hy-mt2-lite",
"name": "Tencent Hy-MT2-Lite",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8000,
"output": 4000
},
"cost": {
"input": 0.044,
"output": 0.177
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/tencent/hy-mt2-lite\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"tencent/hy-mt2-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/hy3": {
"id": "tencent/hy3",
"name": "Hy3",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.14,
"output": 0.58,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/tencent/hy3\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"tencent/hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/hy4-preview": {
"id": "tencent/hy4-preview",
"name": "Tencent Hy4 Preview",
"description": "A next-generation productivity model with significantly enhanced Agent and complex task execution capabilities.",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-08-28",
"last_updated": "2026-08-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1024000,
"output": 64000
},
"cost": {
"input": 0.834,
"output": 2.501,
"cache_read": 0.042
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/tencent/hy4-preview\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"tencent/hy4-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/hy-mt2-plus": {
"id": "tencent/hy-mt2-plus",
"name": "Tencent Hy-MT2-Plus",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8000,
"output": 4000
},
"cost": {
"input": 0.074,
"output": 0.295
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/tencent/hy-mt2-plus\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"tencent/hy-mt2-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/hy-mt2-pro": {
"id": "tencent/hy-mt2-pro",
"name": "Tencent Hy-MT2-Pro",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8000,
"output": 4000
},
"cost": {
"input": 0.074,
"output": 0.295
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/tencent/hy-mt2-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"tencent/hy-mt2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.7": {
"id": "zai/glm-4.7",
"name": "GLM 4.7",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 120000
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-4.7\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.5-air": {
"id": "zai/glm-4.5-air",
"name": "GLM 4.5 Air",
"description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
"family": "glm-air",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 96000
},
"cost": {
"input": 0.2,
"output": 1.1,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-4.5-air\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.5-air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.6": {
"id": "zai/glm-4.6",
"name": "GLM 4.6",
"description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 96000
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-4.6\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.2": {
"id": "zai/glm-5.2",
"name": "GLM 5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.8,
"output": 2.55,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-5.2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.3-fast": {
"id": "zai/glm-5.3-fast",
"name": "GLM 5.3 Fast",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 2.1,
"output": 6.6,
"cache_read": 0.21
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-5.3-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.2-fast": {
"id": "zai/glm-5.2-fast",
"name": "GLM 5.2 Fast",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2.1,
"output": 6.6,
"cache_read": 0.21
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-5.2-fast\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.2-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.3-flash": {
"id": "zai/glm-5.3-flash",
"name": "GLM 5.3 Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131000
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-5.3-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.5": {
"id": "zai/glm-4.5",
"name": "GLM 4.5",
"description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 96000
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-4.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.5v": {
"id": "zai/glm-4.5v",
"name": "GLM 4.5V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08",
"release_date": "2025-08-11",
"last_updated": "2025-08-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 66000,
"output": 16000
},
"cost": {
"input": 0.6,
"output": 1.8,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-4.5v\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.5v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.7-flashx": {
"id": "zai/glm-4.7-flashx",
"name": "GLM 4.7 FlashX",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 0.06,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-4.7-flashx\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.7-flashx\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5": {
"id": "zai/glm-5",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202800,
"output": 131100
},
"cost": {
"input": 1,
"output": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.1": {
"id": "zai/glm-5.1",
"name": "GLM 5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202800,
"output": 64000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-5.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5-turbo": {
"id": "zai/glm-5-turbo",
"name": "GLM 5 Turbo",
"description": "Faster GLM-5 lane for coding agents that need lower latency",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 202800,
"output": 131100
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-5-turbo\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5.3": {
"id": "zai/glm-5.3",
"name": "GLM 5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-5.3\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-5v-turbo": {
"id": "zai/glm-5v-turbo",
"name": "GLM 5V Turbo",
"description": "Fast GLM vision model for screenshots, documents, and multimodal agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-04-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-5v-turbo\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-5v-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/glm-4.7-flash": {
"id": "zai/glm-4.7-flash",
"name": "GLM 4.7 Flash",
"description": "Budget GLM lane for fast coding help, routing, and everyday automation",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131000
},
"cost": {
"input": 0.07,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/zai/glm-4.7-flash\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"zai/glm-4.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/devstral-2": {
"id": "mistral/devstral-2",
"name": "Devstral 2",
"description": "Mistral coding agent model for repository tasks and software engineering workflows",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-12-09",
"last_updated": "2025-12-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.4,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/devstral-2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/devstral-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-medium": {
"id": "mistral/mistral-medium",
"name": "Mistral Medium 3.1",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-medium",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-05-07",
"last_updated": "2025-05-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 64000
},
"cost": {
"input": 0.4,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/mistral-medium\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-medium\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/devstral-small-2": {
"id": "mistral/devstral-small-2",
"name": "Devstral Small 2",
"description": "Mistral coding agent model for repository tasks and software engineering workflows",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-12-09",
"last_updated": "2025-05-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.1,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/devstral-small-2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/devstral-small-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-embed": {
"id": "mistral/mistral-embed",
"name": "Mistral Embed",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "mistral-embed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2023-12-11",
"last_updated": "2023-12-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/mistral-embed\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-embed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-large-3": {
"id": "mistral/mistral-large-3",
"name": "Mistral Large 3",
"description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
"family": "mistral-large",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.5,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/mistral-large-3\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-large-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-nemo": {
"id": "mistral/mistral-nemo",
"name": "Mistral Nemo",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-nemo",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-07-18",
"last_updated": "2024-07-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.15,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/mistral-nemo\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-nemo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/codestral-embed": {
"id": "mistral/codestral-embed",
"name": "Codestral Embed",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "codestral-embed",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2025-05-28",
"last_updated": "2025-05-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 1536
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/codestral-embed\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/codestral-embed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-small": {
"id": "mistral/mistral-small",
"name": "Mistral Small (latest)",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2024-09-17",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 4000
},
"cost": {
"input": 0.1,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/mistral-small\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/ministral-14b": {
"id": "mistral/ministral-14b",
"name": "Ministral 14B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-12-02",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.2,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/ministral-14b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/ministral-14b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-medium-3.5": {
"id": "mistral/mistral-medium-3.5",
"name": "Mistral Medium Latest",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-medium",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-04-29",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 1.5,
"output": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/mistral-medium-3.5\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-medium-3.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/codestral": {
"id": "mistral/codestral",
"name": "Codestral (latest)",
"description": "Mistral code model for completions, refactors, and developer IDE workflows",
"family": "codestral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-05-29",
"last_updated": "2025-01-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 4096
},
"cost": {
"input": 0.3,
"output": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/codestral\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/codestral\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/ministral-8b": {
"id": "mistral/ministral-8b",
"name": "Ministral 8B (latest)",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-10-01",
"last_updated": "2024-10-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.1,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/ministral-8b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/ministral-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/ministral-3b": {
"id": "mistral/ministral-3b",
"name": "Ministral 3B (latest)",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-10-01",
"last_updated": "2024-10-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.04,
"output": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/ministral-3b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/ministral-3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/pixtral-12b": {
"id": "mistral/pixtral-12b",
"name": "Pixtral 12B",
"description": "Mistral vision-language model for image understanding and multimodal chat",
"family": "pixtral",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2024-09-01",
"last_updated": "2024-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.15,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/mistral/pixtral-12b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral/pixtral-12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/pplx-embed-v1-4b": {
"id": "perplexity/pplx-embed-v1-4b",
"name": "Embed v1 4b",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-02-26",
"last_updated": "2026-02-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/perplexity/pplx-embed-v1-4b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/pplx-embed-v1-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/pplx-embed-v1-0.6b": {
"id": "perplexity/pplx-embed-v1-0.6b",
"name": "Embed v1 0.6b",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "v0",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2026-02-26",
"last_updated": "2026-02-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/perplexity/pplx-embed-v1-0.6b\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/pplx-embed-v1-0.6b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/sonar": {
"id": "perplexity/sonar",
"name": "Sonar",
"description": "Sonar search model for current answers, retrieval, and citation-backed chat",
"family": "sonar",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-02",
"release_date": "2025-02-19",
"last_updated": "2025-02-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 127000,
"output": 8000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/perplexity/sonar\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/sonar\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/sonar-reasoning-pro": {
"id": "perplexity/sonar-reasoning-pro",
"name": "Sonar Reasoning Pro",
"description": "Web-grounded reasoning model for multi-step research and cited answers",
"family": "sonar-reasoning",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-02-19",
"last_updated": "2025-02-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 127000,
"output": 8000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/perplexity/sonar-reasoning-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/sonar-reasoning-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/sonar-pro": {
"id": "perplexity/sonar-pro",
"name": "Sonar Pro",
"description": "Advanced Sonar search model for deeper research and cited synthesis",
"family": "sonar-pro",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2025-02-19",
"last_updated": "2025-02-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8000
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/perplexity/sonar-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/sonar-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"recraft/recraft-v4-pro": {
"id": "recraft/recraft-v4-pro",
"name": "Recraft V4 Pro",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "recraft",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-02-17",
"last_updated": "2026-02-17",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/recraft/recraft-v4-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"recraft/recraft-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"recraft/recraft-v4.1-utility": {
"id": "recraft/recraft-v4.1-utility",
"name": "Recraft V4.1 Utility",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "recraft",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-05-14",
"last_updated": "2026-05-14",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/recraft/recraft-v4.1-utility\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"recraft/recraft-v4.1-utility\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"recraft/recraft-v4.1-utility-pro": {
"id": "recraft/recraft-v4.1-utility-pro",
"name": "Recraft V4.1 Utility Pro",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "recraft",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-05-14",
"last_updated": "2026-05-14",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/recraft/recraft-v4.1-utility-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"recraft/recraft-v4.1-utility-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"recraft/recraft-v2": {
"id": "recraft/recraft-v2",
"name": "Recraft V2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "recraft",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-03-13",
"last_updated": "2024-03",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 512,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/recraft/recraft-v2\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"recraft/recraft-v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"recraft/recraft-v3": {
"id": "recraft/recraft-v3",
"name": "Recraft V3",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "recraft",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2024-10-30",
"last_updated": "2024-10",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 512,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/recraft/recraft-v3\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"recraft/recraft-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"recraft/recraft-v4.1-pro": {
"id": "recraft/recraft-v4.1-pro",
"name": "Recraft V4.1 Pro",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "recraft",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-05-14",
"last_updated": "2026-05-14",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/recraft/recraft-v4.1-pro\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"recraft/recraft-v4.1-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"recraft/recraft-v4": {
"id": "recraft/recraft-v4",
"name": "Recraft V4",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "recraft",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-02-17",
"last_updated": "2026-02-17",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/recraft/recraft-v4\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"recraft/recraft-v4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"recraft/recraft-v4.1": {
"id": "recraft/recraft-v4.1",
"name": "Recraft V4.1",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "recraft",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-05-14",
"last_updated": "2026-05-14",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"vercel/recraft/recraft-v4.1\", apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"AI_GATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"recraft/recraft-v4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"qvac": {
"id": "qvac",
"name": "QVAC",
"baseURL": "",
"npm": "@qvac/ai-sdk-provider",
"swiftDriver": "openaiChat",
"env": [
"QVAC_API_KEY"
],
"doc": "https://www.npmjs.com/package/@qvac/ai-sdk-provider",
"modelCount": 9,
"models": {
"qwen3.5-9b": {
"id": "qwen3.5-9b",
"name": "Qwen3.5 9B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qvac/qwen3.5-9b\", apiKey: processEnvironment[\"QVAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"QVAC_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma4-31b": {
"id": "gemma4-31b",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qvac/gemma4-31b\", apiKey: processEnvironment[\"QVAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"QVAC_API_KEY\"]\n)\nlet session = provider.model(\"gemma4-31b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-20b": {
"id": "gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qvac/gpt-oss-20b\", apiKey: processEnvironment[\"QVAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"QVAC_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-27b": {
"id": "qwen3.6-27b",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qvac/qwen3.6-27b\", apiKey: processEnvironment[\"QVAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"QVAC_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-0.8b": {
"id": "qwen3.5-0.8b",
"name": "Qwen3.5 0.8B",
"description": "Qwen instruction model for multilingual chat and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-11-01",
"last_updated": "2025-11-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qvac/qwen3.5-0.8b\", apiKey: processEnvironment[\"QVAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"QVAC_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-0.8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-35b-a3b": {
"id": "qwen3.6-35b-a3b",
"name": "Qwen3.6 35B-A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qvac/qwen3.6-35b-a3b\", apiKey: processEnvironment[\"QVAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"QVAC_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-4b": {
"id": "qwen3.5-4b",
"name": "Qwen3.5 4B",
"description": "Qwen instruction model for multilingual chat and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-11-01",
"last_updated": "2025-11-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qvac/qwen3.5-4b\", apiKey: processEnvironment[\"QVAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"QVAC_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-2b": {
"id": "qwen3.5-2b",
"name": "Qwen3.5 2B",
"description": "Qwen instruction model for multilingual chat and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-11-01",
"last_updated": "2025-11-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qvac/qwen3.5-2b\", apiKey: processEnvironment[\"QVAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"QVAC_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-2b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"qvac/gpt-oss-120b\", apiKey: processEnvironment[\"QVAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"QVAC_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"wandb": {
"id": "wandb",
"name": "Weights & Biases",
"baseURL": "https://api.inference.wandb.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"WANDB_API_KEY"
],
"doc": "https://docs.wandb.ai/guides/integrations/inference/",
"modelCount": 27,
"models": {
"deepseek-ai/DeepSeek-V4-Flash": {
"id": "deepseek-ai/DeepSeek-V4-Flash",
"name": "DeepSeek V4 Flash",
"description": "DeepSeek V4-Flash is an MoE model with 1M context length great for coding, reasoning, and agentic workloads.",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/deepseek-ai/DeepSeek-V4-Flash\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Flash-0731": {
"id": "deepseek-ai/DeepSeek-V4-Flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "DeepSeek V4-Flash-0731 is an MoE model great for coding, reasoning, and agentic workloads.",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.13,
"output": 0.28,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/deepseek-ai/DeepSeek-V4-Flash-0731\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.1": {
"id": "deepseek-ai/DeepSeek-V3.1",
"name": "DeepSeek V3.1",
"description": "A large hybrid model that supports both thinking and non-thinking modes via prompt templates.",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 161000,
"output": 161000
},
"cost": {
"input": 0.55,
"output": 1.65,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/deepseek-ai/DeepSeek-V3.1\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro-0813": {
"id": "deepseek-ai/DeepSeek-V4-Pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4-Pro-0813 is a 1.6T-parameter MoE model excelling at advanced reasoning, coding, and complex agentic workloads.",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.31,
"output": 3.96,
"cache_read": 0.044
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/deepseek-ai/DeepSeek-V4-Pro-0813\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro": {
"id": "deepseek-ai/DeepSeek-V4-Pro",
"name": "DeepSeek V4 Pro",
"description": "DeepSeek V4-Pro is a 1.6T-parameter MoE model with 49B active parameters excelling at advanced reasoning, coding, and complex agentic workloads.",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.15,
"output": 2.55,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/deepseek-ai/DeepSeek-V4-Pro\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B": {
"id": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B",
"name": "Nemotron 3 Ultra",
"description": "Nemotron 3 Ultra is a powerful MoE model designed for long-running agents across coding, deep research, and enterprise automation.",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.75,
"output": 2.75,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B": {
"id": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B",
"name": "Nemotron 3.5 Lightning",
"description": "Nemotron 3.5 Lightning is an MoE model built for fast, reliable agentic tasks across use cases such as financial services, cybersecurity, telecom, and retail.",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-11",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.1,
"output": 0.25,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"OpenPipe/Qwen3-14B-Instruct": {
"id": "OpenPipe/Qwen3-14B-Instruct",
"name": "Qwen3 14B Instruct",
"description": "An efficient multilingual, dense, instruction-tuned model, optimized by OpenPipe for building agents with finetuning.",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-29",
"last_updated": "2025-04-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.22,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/OpenPipe/Qwen3-14B-Instruct\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"OpenPipe/Qwen3-14B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31B-it": {
"id": "google/gemma-4-31B-it",
"name": "Gemma 4 31B",
"description": "Gemma 4 31B Dense is designed for advanced reasoning, agentic workflows, and longer context and is natively trained on 140+ languages.",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.1,
"output": 0.34,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/google/gemma-4-31B-it\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2": {
"id": "zai-org/GLM-5.2",
"name": "GLM 5.2",
"description": "GLM-5.2 is a Mixture-of-Experts language model featuring 40 billion activated parameters and a total of 744 billion parameters.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-16",
"last_updated": "2026-06-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.76,
"output": 2.42,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/zai-org/GLM-5.2\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.3-Flash": {
"id": "zai-org/GLM-5.3-Flash",
"name": "GLM 5.3 Flash",
"description": "GLM-5.3-Flash is a natively multimodal model with 320B total parameters and 18B active parameters.",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/zai-org/GLM-5.3-Flash\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.3-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-30B-A3B-Instruct-2507": {
"id": "Qwen/Qwen3-30B-A3B-Instruct-2507",
"name": "Qwen3 30B A3B Instruct 2507",
"description": "Qwen3-30B-A3B-Instruct-2507 is a 30.5B MoE instruction-tuned model with enhanced reasoning, coding, and long-context understanding.",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-29",
"last_updated": "2025-07-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/Qwen/Qwen3-30B-A3B-Instruct-2507\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-30B-A3B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.8-27B": {
"id": "Qwen/Qwen3.8-27B",
"name": "Qwen3.8 27B",
"description": "Qwen3.8-27B is a dense multimodal model suited for coding, research, vision, and long-running agent tasks.",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.4,
"output": 3,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/Qwen/Qwen3.8-27B\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.8-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-35B-A3B": {
"id": "Qwen/Qwen3.5-35B-A3B",
"name": "Qwen3.5-35B-A3B",
"description": "Qwen3.5-35B-A3B is an open-weights multimodal MoE model built for efficient, high-throughput inference across chat, reasoning, and agentic tasks.",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-24",
"last_updated": "2026-02-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.25,
"output": 1.25,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/Qwen/Qwen3.5-35B-A3B\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-27B": {
"id": "Qwen/Qwen3.6-27B",
"name": "Qwen3.6 27B",
"description": "Qwen3.6-27B is a 27B dense multimodal model with 262K context built for flagship-level agentic coding.",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 3.6,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/Qwen/Qwen3.6-27B\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-35B-A3B": {
"id": "Qwen/Qwen3.6-35B-A3B",
"name": "Qwen3.6 35B A3B",
"description": "Qwen3.6-35B-A3B is an MoE multimodal model with 262K context optimized for agentic coding workflows.",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-15",
"last_updated": "2026-04-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.25,
"output": 1.25,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/Qwen/Qwen3.6-35B-A3B\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ibm-granite/granite-4.1-8b": {
"id": "ibm-granite/granite-4.1-8b",
"name": "Granite 4.1 8B",
"description": "Granite 4.1 8B is a long-context instruct model capable of enhanced tool calling, instruction following, and chat capabilities.",
"family": "granite",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-29",
"last_updated": "2026-04-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.1,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/ibm-granite/granite-4.1-8b\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"ibm-granite/granite-4.1-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ibm-granite/granite-4.2-8b": {
"id": "ibm-granite/granite-4.2-8b",
"name": "Granite 4.2 8B",
"description": "Granite 4.2 8B is an instruct model capable of enhanced tool calling, instruction following, and chat capabilities.",
"family": "granite",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-24",
"last_updated": "2026-08-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.15,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/ibm-granite/granite-4.2-8b\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"ibm-granite/granite-4.2-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M3": {
"id": "MiniMaxAI/MiniMax-M3",
"name": "MiniMax M3",
"description": "MiniMax M3 is a multimodal MoE model with 23B active parameters optimized for coding and agentic workflows.",
"family": "minimax-m3",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.23,
"output": 0.96,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/MiniMaxAI/MiniMax-M3\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.1-8B-Instruct": {
"id": "meta-llama/Llama-3.1-8B-Instruct",
"name": "Llama 3.1 8B",
"description": "Efficient conversational model optimized for responsive multilingual chatbot interactions.",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.22,
"output": 0.22,
"cache_read": 0.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/meta-llama/Llama-3.1-8B-Instruct\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.1-8B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.1-70B-Instruct": {
"id": "meta-llama/Llama-3.1-70B-Instruct",
"name": "Llama 3.1 70B",
"description": "Efficient conversational model optimized for responsive multilingual chatbot interactions.",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.8,
"output": 0.8,
"cache_read": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/meta-llama/Llama-3.1-70B-Instruct\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.1-70B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.3-70B-Instruct": {
"id": "meta-llama/Llama-3.3-70B-Instruct",
"name": "Llama 3.3 70B",
"description": "Multilingual model excelling in conversational tasks, detailed instruction-following, and coding.",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-01",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.71,
"output": 0.71,
"cache_read": 0.71
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/meta-llama/Llama-3.3-70B-Instruct\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.3-70B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "gpt-oss-20b",
"description": "Lower latency Mixture-of-Experts model trained on OpenAI's Harmony response format with reasoning capabilities.",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.03,
"output": 0.13,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/openai/gpt-oss-20b\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "gpt-oss-120b",
"description": "Efficient Mixture-of-Experts model designed for high-reasoning, agentic and general-purpose use cases.",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.03,
"output": 0.17,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/openai/gpt-oss-120b\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.7-Code": {
"id": "moonshotai/Kimi-K2.7-Code",
"name": "Kimi K2.7 Code",
"description": "Kimi K2.7 Code is a 1T-parameter MoE model with 32B active parameters purpose-built for long-horizon agentic coding and software engineering.",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.71,
"output": 3.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/moonshotai/Kimi-K2.7-Code\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.7-Code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6": {
"id": "moonshotai/Kimi-K2.6",
"name": "Kimi K2.6",
"description": "Kimi K2.6 is a multimodal Mixture-of-Experts language model featuring 32 billion activated parameters and a total of 1 trillion parameters.",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-20",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.65,
"output": 3.41,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"JetBrains/Mellum2-12B-A2.5B-Instruct": {
"id": "JetBrains/Mellum2-12B-A2.5B-Instruct",
"name": "Mellum2 12B A2.5B",
"description": "Mellum2-12B-A2.5B-Instruct is a fast MoE model with 131K context built for coding, tool use, and low-latency AI workflows.",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.1,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wandb/JetBrains/Mellum2-12B-A2.5B-Instruct\", apiKey: processEnvironment[\"WANDB_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.wandb.ai/v1\")!,\n apiKey: processEnvironment[\"WANDB_API_KEY\"]\n)\nlet session = provider.model(\"JetBrains/Mellum2-12B-A2.5B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"friendli": {
"id": "friendli",
"name": "Friendli",
"baseURL": "https://api.friendli.ai/serverless/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"FRIENDLI_TOKEN"
],
"doc": "https://friendli.ai/docs/guides/serverless_endpoints/introduction",
"modelCount": 6,
"models": {
"deepseek-ai/DeepSeek-V3.2": {
"id": "deepseek-ai/DeepSeek-V3.2",
"name": "DeepSeek-V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 0.5,
"output": 1.5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"friendli/deepseek-ai/DeepSeek-V3.2\", apiKey: processEnvironment[\"FRIENDLI_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.friendli.ai/serverless/v1\")!,\n apiKey: processEnvironment[\"FRIENDLI_TOKEN\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31B-it": {
"id": "google/gemma-4-31B-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.14,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"friendli/google/gemma-4-31B-it\", apiKey: processEnvironment[\"FRIENDLI_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.friendli.ai/serverless/v1\")!,\n apiKey: processEnvironment[\"FRIENDLI_TOKEN\"]\n)\nlet session = provider.model(\"google/gemma-4-31B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.1": {
"id": "zai-org/GLM-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 202752
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"friendli/zai-org/GLM-5.1\", apiKey: processEnvironment[\"FRIENDLI_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.friendli.ai/serverless/v1\")!,\n apiKey: processEnvironment[\"FRIENDLI_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.3": {
"id": "zai-org/GLM-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": -1,
"max": 1048576
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"friendli/zai-org/GLM-5.3\", apiKey: processEnvironment[\"FRIENDLI_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.friendli.ai/serverless/v1\")!,\n apiKey: processEnvironment[\"FRIENDLI_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2": {
"id": "zai-org/GLM-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"friendli/zai-org/GLM-5.2\", apiKey: processEnvironment[\"FRIENDLI_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.friendli.ai/serverless/v1\")!,\n apiKey: processEnvironment[\"FRIENDLI_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M2.5": {
"id": "MiniMaxAI/MiniMax-M2.5",
"name": "MiniMax-M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 196608
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"friendli/MiniMaxAI/MiniMax-M2.5\", apiKey: processEnvironment[\"FRIENDLI_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.friendli.ai/serverless/v1\")!,\n apiKey: processEnvironment[\"FRIENDLI_TOKEN\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"tokenrouter": {
"id": "tokenrouter",
"name": "TokenRouter",
"baseURL": "https://api.tokenrouter.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"TOKENROUTER_API_KEY"
],
"doc": "https://www.tokenrouter.com/docs/tokenrouter-feature-guide/",
"modelCount": 1,
"models": {
"z-ai/glm-5.3-free": {
"id": "z-ai/glm-5.3-free",
"name": "GLM-5.3 (free)",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tokenrouter/z-ai/glm-5.3-free\", apiKey: processEnvironment[\"TOKENROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tokenrouter.com/v1\")!,\n apiKey: processEnvironment[\"TOKENROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.3-free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"thinkingmachines": {
"id": "thinkingmachines",
"name": "Thinking Machines",
"baseURL": "https://tinker.thinkingmachines.dev/services/tinker-prod/anthropic/api/v1",
"npm": "@ai-sdk/anthropic",
"swiftDriver": "anthropicMessages",
"env": [
"TINKER_API_KEY"
],
"doc": "https://tinker-docs.thinkingmachines.ai/tinker/compatible-apis/anthropic/",
"modelCount": 2,
"models": {
"thinkingmachines/Inkling": {
"id": "thinkingmachines/Inkling",
"name": "Inkling",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 1.87,
"output": 4.68,
"cache_read": 0.374
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"thinkingmachines/thinkingmachines/Inkling\", apiKey: processEnvironment[\"TINKER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://tinker.thinkingmachines.dev/services/tinker-prod/anthropic/api/v1\")!,\n apiKey: processEnvironment[\"TINKER_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/Inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/Inkling:peft:262144": {
"id": "thinkingmachines/Inkling:peft:262144",
"name": "Inkling (256K)",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 3.74,
"output": 9.36,
"cache_read": 0.748
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"thinkingmachines/thinkingmachines/Inkling:peft:262144\", apiKey: processEnvironment[\"TINKER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://tinker.thinkingmachines.dev/services/tinker-prod/anthropic/api/v1\")!,\n apiKey: processEnvironment[\"TINKER_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/Inkling:peft:262144\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"standardcompute": {
"id": "standardcompute",
"name": "Standard Compute",
"baseURL": "https://api.stdcmpt.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"STANDARDCOMPUTE_API_KEY"
],
"doc": "https://standardcompute.com/models",
"modelCount": 1,
"models": {
"standardcompute": {
"id": "standardcompute",
"name": "Standard Compute",
"description": "Flat-rate smart-routing gateway: one model id, each request routed across a curated catalog of 1M-context models (DeepSeek, GLM, MiniMax, Qwen, GPT-5.6, Claude 5, Gemini 2.5, Kimi) or pinned to a user-selected model",
"family": "auto",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_details"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-01",
"last_updated": "2026-08-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 24576
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"standardcompute/standardcompute\", apiKey: processEnvironment[\"STANDARDCOMPUTE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.stdcmpt.com/v1\")!,\n apiKey: processEnvironment[\"STANDARDCOMPUTE_API_KEY\"]\n)\nlet session = provider.model(\"standardcompute\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"tensorx": {
"id": "tensorx",
"name": "TensorX",
"baseURL": "https://api.tensorx.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"TENSORX_API_KEY"
],
"doc": "https://docs.tensorx.ai/",
"modelCount": 25,
"models": {
"qwen/qwen3.5-9b": {
"id": "qwen/qwen3.5-9b",
"name": "Qwen3.5 9B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 0.2,
"cache_read": 0.0375,
"cache_write": 0.1875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/qwen/qwen3.5-9b\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-30b-a3b-instruct": {
"id": "qwen/qwen3-coder-30b-a3b-instruct",
"name": "Qwen3-Coder 30B-A3B Instruct",
"description": "Smaller Qwen coder for efficient local agents and repo-level fixes",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 65536
},
"cost": {
"input": 0.06,
"output": 0.25,
"cache_read": 0.015,
"cache_write": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/qwen/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b-2507": {
"id": "qwen/qwen3-235b-a22b-2507",
"name": "Qwen3 235B-A22B-2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-06-30",
"release_date": "2025-07-21",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131000,
"output": 262144
},
"cost": {
"input": 0.072,
"output": 0.464,
"cache_read": 0.018,
"cache_write": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/qwen/qwen3-235b-a22b-2507\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-122b-a10b": {
"id": "qwen/qwen3.5-122b-a10b",
"name": "Qwen3.5 122B-A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.5,
"output": 3.5,
"cache_read": 0.125,
"cache_write": 0.625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/qwen/qwen3.5-122b-a10b\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-122b-a10b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-235b-a22b-instruct": {
"id": "qwen/qwen3-vl-235b-a22b-instruct",
"name": "Qwen3 VL 235B-A22B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131000,
"output": 131072
},
"cost": {
"input": 0.21,
"output": 1.9,
"cache_read": 0.0525,
"cache_write": 0.2625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/qwen/qwen3-vl-235b-a22b-instruct\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-235b-a22b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.5": {
"id": "minimax/minimax-m2.5",
"name": "MiniMax-M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.075,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/minimax/minimax-m2.5\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m3": {
"id": "minimax/minimax-m3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.4,
"output": 2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/minimax/minimax-m3\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/nemotron-3-super-120b-a12b": {
"id": "nvidia/nemotron-3-super-120b-a12b",
"name": "Nemotron 3 Super 120B A12B",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.3,
"output": 0.9,
"cache_read": 0.075,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/nvidia/nemotron-3-super-120b-a12b\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-super-120b-a12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-chat-v3.1": {
"id": "deepseek/deepseek-chat-v3.1",
"name": "DeepSeek Chat V3.1",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-11",
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 164000,
"output": 163840
},
"cost": {
"input": 0.2,
"output": 0.8,
"cache_read": 0.05,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/deepseek/deepseek-chat-v3.1\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-chat-v3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-0731": {
"id": "deepseek/deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.25,
"output": 0.3,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/deepseek/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash": {
"id": "deepseek/deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.15,
"output": 0.3,
"cache_read": 0.0375,
"cache_write": 0.1875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/deepseek/deepseek-v4-flash\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-r1-0528": {
"id": "deepseek/deepseek-r1-0528",
"name": "DeepSeek R1-0528",
"description": "Classic open reasoning model for transparent math, coding, and deliberate problem solving",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-05-28",
"last_updated": "2025-05-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 164000,
"output": 8192
},
"cost": {
"input": 0.66,
"output": 2.6,
"cache_read": 0.165,
"cache_write": 0.825
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/deepseek/deepseek-r1-0528\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-r1-0528\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v3.2": {
"id": "deepseek/deepseek-v3.2",
"name": "DeepSeek V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 0.3,
"output": 0.5,
"cache_read": 0.075,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/deepseek/deepseek-v3.2\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro": {
"id": "deepseek/deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 1.75,
"output": 3.5,
"cache_read": 0.4375,
"cache_write": 2.185
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/deepseek/deepseek-v4-pro\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.04,
"output": 0.2,
"cache_read": 0.01,
"cache_write": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/openai/gpt-oss-120b\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.6": {
"id": "moonshotai/kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1,
"output": 4,
"cache_read": 0.25,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/moonshotai/kimi-k2.6\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.7-code": {
"id": "moonshotai/kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.25,
"output": 4.5,
"cache_read": 0.3125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/moonshotai/kimi-k2.7-code\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k3": {
"id": "moonshotai/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/moonshotai/kimi-k3\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.5": {
"id": "moonshotai/kimi-k2.5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.5,
"output": 2.8,
"cache_read": 0.125,
"cache_write": 0.625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/moonshotai/kimi-k2.5\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-4.7": {
"id": "z-ai/glm-4.7",
"name": "GLM-4.7",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 200000
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.15,
"cache_write": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/z-ai/glm-4.7\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.2": {
"id": "z-ai/glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.5,
"output": 4.5,
"cache_read": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/z-ai/glm-5.2\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5": {
"id": "z-ai/glm-5",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 202752
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.25,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/z-ai/glm-5\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.1": {
"id": "z-ai/glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 202752
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.35,
"cache_write": 1.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/z-ai/glm-5.1\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5-turbo": {
"id": "z-ai/glm-5-turbo",
"name": "GLM-5-Turbo",
"description": "Faster GLM-5 lane for coding agents that need lower latency",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.3,
"cache_write": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/z-ai/glm-5-turbo\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5v-turbo": {
"id": "z-ai/glm-5v-turbo",
"name": "GLM-5V-Turbo",
"description": "Fast GLM vision model for screenshots, documents, and multimodal agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-04-01",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.3,
"cache_write": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"tensorx/z-ai/glm-5v-turbo\", apiKey: processEnvironment[\"TENSORX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.tensorx.ai/v1\")!,\n apiKey: processEnvironment[\"TENSORX_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5v-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"meta": {
"id": "meta",
"name": "Meta",
"baseURL": "https://api.meta.ai/v1",
"npm": "@ai-sdk/openai",
"swiftDriver": "openaiChat",
"env": [
"META_MODEL_API_KEY"
],
"doc": "https://dev.meta.ai/docs",
"modelCount": 5,
"models": {
"muse-spark-1.3": {
"id": "muse-spark-1.3",
"name": "Muse Spark 1.3",
"description": "Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It improves long-horizon agent collaboration, instruction following, and coding efficiency relative to Muse Spark 1.2.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"meta/muse-spark-1.3\", apiKey: processEnvironment[\"META_MODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.meta.ai/v1\")!,\n apiKey: processEnvironment[\"META_MODEL_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1.2": {
"id": "muse-spark-1.2",
"name": "Muse Spark 1.2",
"description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"meta/muse-spark-1.2\", apiKey: processEnvironment[\"META_MODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.meta.ai/v1\")!,\n apiKey: processEnvironment[\"META_MODEL_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1.2-contributor": {
"id": "muse-spark-1.2-contributor",
"name": "Muse Spark 1.2 Contributor",
"description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.002
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"meta/muse-spark-1.2-contributor\", apiKey: processEnvironment[\"META_MODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.meta.ai/v1\")!,\n apiKey: processEnvironment[\"META_MODEL_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1.2-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1.3-contributor": {
"id": "muse-spark-1.3-contributor",
"name": "Muse Spark 1.3 Contributor",
"description": "Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It improves long-horizon agent collaboration, instruction following, and coding efficiency relative to Muse Spark 1.2.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.002
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"meta/muse-spark-1.3-contributor\", apiKey: processEnvironment[\"META_MODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.meta.ai/v1\")!,\n apiKey: processEnvironment[\"META_MODEL_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1.3-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1.1": {
"id": "muse-spark-1.1",
"name": "Muse Spark 1.1",
"description": "Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-08",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"meta/muse-spark-1.1\", apiKey: processEnvironment[\"META_MODEL_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.meta.ai/v1\")!,\n apiKey: processEnvironment[\"META_MODEL_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"venice": {
"id": "venice",
"name": "Venice AI",
"baseURL": "",
"npm": "venice-ai-sdk-provider",
"swiftDriver": "openaiChat",
"env": [
"VENICE_API_KEY"
],
"doc": "https://docs.venice.ai",
"modelCount": 107,
"models": {
"google-gemma-3-27b-it": {
"id": "google-gemma-3-27b-it",
"name": "Google Gemma 3 27B Instruct",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-11-04",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 198000,
"output": 16384
},
"cost": {
"input": 0.12,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/google-gemma-3-27b-it\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"google-gemma-3-27b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org-glm-5-2": {
"id": "zai-org-glm-5-2",
"name": "GLM 5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-16",
"last_updated": "2026-06-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/zai-org-glm-5-2\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"zai-org-glm-5-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3.6,
"output": 18,
"cache_read": 0.36,
"cache_write": 4.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-sonnet-4-6\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731-fast": {
"id": "deepseek-v4-flash-0731-fast",
"name": "DeepSeek V4 Flash 0731 Fast",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-08-09",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.35,
"output": 0.7,
"cache_read": 0.0875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/deepseek-v4-flash-0731-fast\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-5-9b": {
"id": "qwen3-5-9b",
"name": "Qwen 3.5 9B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-05",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen3-5-9b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-55-pro": {
"id": "openai-gpt-55-pro",
"name": "GPT-5.5 Pro",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-24",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 37.5,
"output": 225
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-55-pro\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-55-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org-glm-4.7-flash": {
"id": "zai-org-glm-4.7-flash",
"name": "GLM 4.7 Flash",
"description": "Budget GLM lane for fast coding help, routing, and everyday automation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-29",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.06,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/zai-org-glm-4.7-flash\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"zai-org-glm-4.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-3-2-24b-instruct": {
"id": "mistral-small-3-2-24b-instruct",
"name": "Mistral Small 3.2 24B Instruct",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-01-15",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0.09375,
"output": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/mistral-small-3-2-24b-instruct\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small-3-2-24b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro-0813": {
"id": "deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "Flagship DeepSeek model for coding, reasoning, and agentic work",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 1.65,
"output": 4.95,
"cache_read": 0.165
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.175,
"output": 0.35,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-7-flash": {
"id": "gemini-3-7-flash",
"name": "Gemini 3.7 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.9375,
"output": 4.6875,
"cache_read": 0.09375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/gemini-3-7-flash\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"venice-uncensored-1-2": {
"id": "venice-uncensored-1-2",
"name": "Venice Uncensored 1.2",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"family": "venice",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-04-01",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.2,
"output": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/venice-uncensored-1-2\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"venice-uncensored-1-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-thinking-2507": {
"id": "qwen3-235b-a22b-thinking-2507",
"name": "Qwen 3 235B A22B Thinking 2507",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"release_date": "2025-04-29",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.45,
"output": 3.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen3-235b-a22b-thinking-2507\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-uncensored": {
"id": "gemma-4-uncensored",
"name": "Gemma 4 Uncensored",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-04-13",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 8192
},
"cost": {
"input": 0.1625,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/gemma-4-uncensored\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-uncensored\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org-glm-5-1": {
"id": "zai-org-glm-5-1",
"name": "GLM 5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 80000
},
"cost": {
"input": 1.54,
"output": 4.84,
"cache_read": 0.286
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/zai-org-glm-5-1\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"zai-org-glm-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-3-6-plus": {
"id": "qwen-3-6-plus",
"name": "Qwen 3.6 Plus Uncensored",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-06",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.625,
"output": 3.75,
"cache_read": 0.0625,
"cache_write": 0.78,
"tiers": [
{
"input": 2.5,
"output": 7.5,
"cache_read": 0.0625,
"cache_write": 0.78,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.0625,
"cache_write": 0.78
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen-3-6-plus\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-3-6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-55": {
"id": "openai-gpt-55",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 131072
},
"cost": {
"input": 6.25,
"output": 37.5,
"cache_read": 0.625,
"tiers": [
{
"input": 12.5,
"output": 56.25,
"cache_read": 1.25,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 12.5,
"output": 56.25,
"cache_read": 1.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-55\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-55\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5-fast": {
"id": "claude-opus-5-fast",
"name": "Claude Opus 5 Fast",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-23",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 12,
"output": 60,
"cache_read": 1.2,
"cache_write": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-opus-5-fast\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-5-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m25": {
"id": "minimax-m25",
"name": "MiniMax M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 198000,
"output": 32768
},
"cost": {
"input": 0.27,
"output": 0.95,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/minimax-m25\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m25\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aion-labs-aion-3-0-mini": {
"id": "aion-labs-aion-3-0-mini",
"name": "Aion 3.0 Mini",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.875,
"output": 1.75,
"cache_read": 0.225
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/aion-labs-aion-3-0-mini\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"aion-labs-aion-3-0-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-5-flash": {
"id": "gemini-3-5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-22",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 1.55,
"output": 9.45,
"cache_read": 0.155,
"cache_write": 0.086
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/gemini-3-5-flash\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5": {
"id": "claude-opus-5",
"name": "Claude Opus 5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-23",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 6,
"output": 30,
"cache_read": 0.6,
"cache_write": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-opus-5\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-52-codex": {
"id": "openai-gpt-52-codex",
"name": "GPT-5.2 Codex",
"description": "Code-specialist GPT for repository edits, reviews, and long-running software agents",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08",
"release_date": "2025-01-15",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 272000,
"output": 65536
},
"cost": {
"input": 2.19,
"output": 17.5,
"cache_read": 0.219
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-52-codex\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-52-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-3-7-max": {
"id": "qwen-3-7-max",
"name": "Qwen 3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-22",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 2.7,
"output": 8.05,
"cache_read": 0.27,
"cache_write": 3.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen-3-7-max\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-3-7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-3-8-max": {
"id": "qwen-3-8-max",
"name": "Qwen 3.8 Max",
"description": "Preview Qwen flagship for million-token multimodal reasoning and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-22",
"last_updated": "2026-07-19",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.3125,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen-3-8-max\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-3-8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-54": {
"id": "openai-gpt-54",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 131072
},
"cost": {
"input": 3.13,
"output": 18.8,
"cache_read": 0.313
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-54\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-54\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-1-flash": {
"id": "deepseek-v4-1-flash",
"name": "DeepSeek V4.1 Flash",
"description": "Fast DeepSeek model for efficient chat, coding help, and agent loops",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.375,
"output": 1.5,
"cache_read": 0.0075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/deepseek-v4-1-flash\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-6-astra": {
"id": "openai-gpt-6-astra",
"name": "GPT-6 Astra",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-04-30",
"release_date": "2026-09-05",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5,
"tiers": [
{
"input": 20,
"output": 75,
"cache_read": 2,
"cache_write": 25,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 20,
"output": 75,
"cache_read": 2,
"cache_write": 25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-6-astra\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-6-astra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai-glm-5-turbo": {
"id": "z-ai-glm-5-turbo",
"name": "GLM 5 Turbo",
"description": "Faster GLM-5 lane for coding agents that need lower latency",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-15",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 32768
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.24
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/z-ai-glm-5-turbo\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"z-ai-glm-5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org-glm-4.6": {
"id": "zai-org-glm-4.6",
"name": "GLM 4.6",
"description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2024-04-01",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 198000,
"output": 16384
},
"cost": {
"input": 0.43,
"output": 1.75,
"cache_read": 0.08
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/zai-org-glm-4.6\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"zai-org-glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5": {
"id": "claude-opus-4-5",
"name": "Claude Opus 4.5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-12-06",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 198000,
"output": 32768
},
"cost": {
"input": 6,
"output": 30,
"cache_read": 0.6,
"cache_write": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-opus-4-5\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash 0423",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.138,
"output": 0.275,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/deepseek-v4-flash\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org-glm-5": {
"id": "zai-org-glm-5",
"name": "GLM 5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-11",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 198000,
"output": 32000
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/zai-org-glm-5\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"zai-org-glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-6-flash": {
"id": "gemini-3-6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-09",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.9375,
"output": 4.6875,
"cache_read": 0.09375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/gemini-3-6-flash\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-56-terra": {
"id": "openai-gpt-56-terra",
"name": "GPT-5.6 Terra",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"cache_write": 3.125,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"cache_write": 6.25,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"cache_write": 6.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-56-terra\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-56-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3-fast-api": {
"id": "kimi-k3-fast-api",
"name": "Kimi K3 Fast",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 4.5,
"output": 22.5,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/kimi-k3-fast-api\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3-fast-api\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-3-8-27b": {
"id": "qwen-3-8-27b",
"name": "Qwen 3.8 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-17",
"last_updated": "2026-08-18",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.45,
"output": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen-3-8-27b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-3-8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-oss-120b": {
"id": "openai-gpt-oss-120b",
"name": "OpenAI GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-11-06",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.07,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-oss-120b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"olafangensan-glm-4.7-flash-heretic": {
"id": "olafangensan-glm-4.7-flash-heretic",
"name": "GLM 4.7 Flash Heretic",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-02-04",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 24000
},
"cost": {
"input": 0.07,
"output": 0.4,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/olafangensan-glm-4.7-flash-heretic\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"olafangensan-glm-4.7-flash-heretic\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-7-code": {
"id": "kimi-k2-7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-13",
"last_updated": "2026-06-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.5,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/kimi-k2-7-code\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aion-labs-aion-3-0": {
"id": "aion-labs-aion-3-0",
"name": "Aion 3.0",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 3.75,
"output": 7.5,
"cache_read": 0.9375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/aion-labs-aion-3-0\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"aion-labs-aion-3-0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.2-3b": {
"id": "llama-3.2-3b",
"name": "Llama 3.2 3B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-10-03",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/llama-3.2-3b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.2-3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-5-397b-a17b": {
"id": "qwen3-5-397b-a17b",
"name": "Qwen 3.5 397B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-16",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.75,
"output": 4.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen3-5-397b-a17b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-2-1-turbo": {
"id": "seed-2-1-turbo",
"name": "Seed 2.1 Turbo",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-28",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.625,
"output": 3.125,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/seed-2-1-turbo\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"seed-2-1-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5-1": {
"id": "claude-fable-5-1",
"name": "Claude Fable 5.1",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-08-29",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 12,
"output": 60,
"cache_read": 0.3,
"cache_write": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-fable-5-1\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-54-pro": {
"id": "openai-gpt-54-pro",
"name": "GPT-5.4 Pro",
"description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 37.5,
"output": 225,
"tiers": [
{
"input": 75,
"output": 337.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 75,
"output": 337.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-54-pro\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-54-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"hermes-3-llama-3.1-405b": {
"id": "hermes-3-llama-3.1-405b",
"name": "Hermes 3 Llama 3.1 405b",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "hermes",
"attachment": false,
"reasoning": false,
"tool_call": false,
"release_date": "2025-09-25",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 1.1,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/hermes-3-llama-3.1-405b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"hermes-3-llama-3.1-405b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"venice-uncensored-role-play": {
"id": "venice-uncensored-role-play",
"name": "Venice Role Play Uncensored",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"family": "venice",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-02-20",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/venice-uncensored-role-play\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"venice-uncensored-role-play\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mercury-2-5": {
"id": "mercury-2-5",
"name": "Mercury 2.5",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"family": "mercury",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-08",
"last_updated": "2026-09-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 260000,
"output": 65536
},
"cost": {
"input": 0.04999999999999999,
"output": 0.18749999999999994,
"cache_read": 0.004999999999999999
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/mercury-2-5\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"mercury-2-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-6": {
"id": "kimi-k2-6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-20",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.5,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/kimi-k2-6\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-6-astra-pro": {
"id": "openai-gpt-6-astra-pro",
"name": "GPT-6 Astra Pro",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-05",
"last_updated": "2026-09-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"output": 128000
},
"cost": {
"input": 12.5,
"output": 62.5,
"cache_read": 1.25,
"cache_write": 15.625,
"tiers": [
{
"input": 25,
"output": 93.75,
"cache_read": 2.5,
"cache_write": 31.25,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 25,
"output": 93.75,
"cache_read": 2.5,
"cache_write": 31.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-6-astra-pro\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-6-astra-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-54-mini": {
"id": "openai-gpt-54-mini",
"name": "GPT-5.4 Mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-27",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.9375,
"output": 5.625,
"cache_read": 0.09375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-54-mini\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-54-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-6-35b-a3b": {
"id": "qwen3-6-35b-a3b",
"name": "Qwen 3.6 35B A3B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-20",
"last_updated": "2026-07-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen3-6-35b-a3b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m3-preview": {
"id": "minimax-m3-preview",
"name": "MiniMax M3 Preview",
"description": "MiniMax multimodal coding model for long-context reasoning and agent tasks",
"family": "minimax-m3",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-06-12",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/minimax-m3-preview\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m3-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-5-35b-a3b": {
"id": "qwen3-5-35b-a3b",
"name": "Qwen 3.5 35B A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-25",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0.3125,
"output": 1.25,
"cache_read": 0.15625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen3-5-35b-a3b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-5-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 6,
"output": 30,
"cache_read": 0.6,
"cache_write": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-opus-4-6\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-5": {
"id": "kimi-k2-5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2024-04",
"release_date": "2026-01-27",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.56,
"output": 3.5,
"cache_read": 0.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/kimi-k2-5\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-5-flash-lite": {
"id": "gemini-3-5-flash-lite",
"name": "Gemini 3.5 Flash-Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-09",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.375,
"output": 3.125,
"cache_read": 0.0375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/gemini-3-5-flash-lite\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-gemma-4-26b-a4b-it": {
"id": "google-gemma-4-26b-a4b-it",
"name": "Google Gemma 4 26B A4B Instruct",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 8192
},
"cost": {
"input": 0.13,
"output": 0.4,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/google-gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"google-gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-53-codex": {
"id": "openai-gpt-53-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-24",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 2.19,
"output": 17.5,
"cache_read": 0.219
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-53-codex\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-53-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-3-8-2-4t-a95b": {
"id": "qwen-3-8-2-4t-a95b",
"name": "Qwen 3.8 2.4T",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.3125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen-3-8-2-4t-a95b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-3-8-2-4t-a95b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7": {
"id": "claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 6,
"output": 30,
"cache_read": 0.6,
"cache_write": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-opus-4-7\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 3.75,
"output": 18.75,
"cache_read": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/kimi-k3\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-56-sol": {
"id": "openai-gpt-56-sol",
"name": "GPT-5.6 Sol",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 12.5,
"cache_read": 0.25,
"cache_write": 3.125,
"tiers": [
{
"input": 5,
"output": 18.75,
"cache_read": 0.5,
"cache_write": 6.25,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 18.75,
"cache_read": 0.5,
"cache_write": 6.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-56-sol\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-56-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3.2": {
"id": "deepseek-v3.2",
"name": "DeepSeek V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-04",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 160000,
"output": 32768
},
"cost": {
"input": 0.33,
"output": 0.48,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/deepseek-v3.2\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xiaomi-mimo-v2-5": {
"id": "xiaomi-mimo-v2-5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-06-11",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 2,
"cache_read": 0.08
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/xiaomi-mimo-v2-5\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"xiaomi-mimo-v2-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-1-pro-preview": {
"id": "gemini-3-1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.5,
"cache_write": 0.5,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"cache_write": 0.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"cache_write": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/gemini-3-1-pro-preview\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-56-terra-pro": {
"id": "openai-gpt-56-terra-pro",
"name": "GPT-5.6 Terra Pro",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"cache_write": 3.125,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"cache_write": 6.25,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"cache_write": 6.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-56-terra-pro\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-56-terra-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-5": {
"id": "grok-4-5",
"name": "Grok 4.5",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-07",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 32000
},
"cost": {
"input": 2.27,
"output": 6.8,
"cache_read": 0.34,
"tiers": [
{
"input": 4.53,
"output": 13.6,
"cache_read": 0.68,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4.53,
"output": 13.6,
"cache_read": 0.68
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/grok-4-5\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-10",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 12,
"output": 60,
"cache_read": 1.2,
"cache_write": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-fable-5\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-52": {
"id": "openai-gpt-52",
"name": "GPT-5.2",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-13",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 272000,
"output": 65536
},
"cost": {
"input": 2.19,
"output": 17.5,
"cache_read": 0.219
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-52\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-52\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8-fast": {
"id": "claude-opus-4-8-fast",
"name": "Claude Opus 4.8 Fast",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 12,
"output": 60,
"cache_read": 1.2,
"cache_write": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-opus-4-8-fast\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-56-luna-pro": {
"id": "openai-gpt-56-luna-pro",
"name": "GPT-5.6 Luna Pro",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"cache_write": 0.3125,
"tiers": [
{
"input": 0.5,
"output": 2.25,
"cache_read": 0.05,
"cache_write": 0.625,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 0.5,
"output": 2.25,
"cache_read": 0.05,
"cache_write": 0.625
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-56-luna-pro\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-56-luna-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai-glm-5-3-flash": {
"id": "z-ai-glm-5-3-flash",
"name": "GLM 5.3 Flash",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-21",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/z-ai-glm-5-3-flash\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"z-ai-glm-5-3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-20": {
"id": "grok-4-20",
"name": "Grok 4.20",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-03-12",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 128000
},
"cost": {
"input": 1.42,
"output": 2.83,
"cache_read": 0.23,
"tiers": [
{
"input": 2.83,
"output": 5.67,
"cache_read": 0.45,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.83,
"output": 5.67,
"cache_read": 0.45
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/grok-4-20\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-20\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google-gemma-4-31b-it": {
"id": "google-gemma-4-31b-it",
"name": "Google Gemma 4 31B Instruct",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-03",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 8192
},
"cost": {
"input": 0.12,
"output": 0.36,
"cache_read": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/google-gemma-4-31b-it\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"google-gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-6-27b": {
"id": "qwen3-6-27b",
"name": "Qwen 3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-24",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.325,
"output": 3.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen3-6-27b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-build-0-1": {
"id": "grok-build-0-1",
"name": "Grok Build 0.1",
"description": "Fast Grok coding model tuned for agentic engineering and iterative edits",
"family": "grok-build",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 1,
"output": 2,
"cache_read": 0.2,
"tiers": [
{
"input": 2,
"output": 4,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2,
"output": 4,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/grok-build-0-1\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"grok-build-0-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-3": {
"id": "grok-4-3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-18",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32000
},
"cost": {
"input": 1.42,
"output": 2.83,
"cache_read": 0.23,
"tiers": [
{
"input": 2.83,
"output": 5.67,
"cache_read": 0.45,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.83,
"output": 5.67,
"cache_read": 0.45
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/grok-4-3\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"inkling": {
"id": "inkling",
"name": "Inkling",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-16",
"last_updated": "2026-07-17",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 5.0625,
"cache_read": 0.2125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/inkling\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-3-8-flash": {
"id": "qwen-3-8-flash",
"name": "Qwen 3.8 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.49,
"cache_read": 0.014
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen-3-8-flash\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-3-8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-480b-a35b-instruct-turbo": {
"id": "qwen3-coder-480b-a35b-instruct-turbo",
"name": "Qwen 3 Coder 480B Turbo",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-01-27",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.35,
"output": 1.5,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen3-coder-480b-a35b-instruct-turbo\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-480b-a35b-instruct-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-4o-2024-11-20": {
"id": "openai-gpt-4o-2024-11-20",
"name": "GPT-4o",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2026-02-28",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 3.125,
"output": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-4o-2024-11-20\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-4o-2024-11-20\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m27": {
"id": "minimax-m27",
"name": "MiniMax M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 198000,
"output": 32768
},
"cost": {
"input": 0.375,
"output": 1.5,
"cache_read": 0.06875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/minimax-m27\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m27\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-next-80b": {
"id": "qwen3-next-80b",
"name": "Qwen 3 Next 80b",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2025-04-29",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0.35,
"output": 1.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen3-next-80b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-next-80b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mercury-2": {
"id": "mercury-2",
"name": "Mercury 2",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"family": "mercury",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-02-20",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 50000
},
"cost": {
"input": 0.3125,
"output": 0.9375,
"cache_read": 0.03125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/mercury-2\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"mercury-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5": {
"id": "claude-sonnet-4-5",
"name": "Claude Sonnet 4.5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-01-15",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 198000,
"output": 64000
},
"cost": {
"input": 3.75,
"output": 18.75,
"cache_read": 0.375,
"cache_write": 4.69
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-sonnet-4-5\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia-nemotron-3-nano-30b-a3b": {
"id": "nvidia-nemotron-3-nano-30b-a3b",
"name": "NVIDIA Nemotron 3 Nano 30B",
"description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
"family": "nemotron",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-01-27",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.075,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/nvidia-nemotron-3-nano-30b-a3b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"nvidia-nemotron-3-nano-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org-glm-4.7": {
"id": "zai-org-glm-4.7",
"name": "GLM 4.7",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-24",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 198000,
"output": 16384
},
"cost": {
"input": 0.55,
"output": 2.65,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/zai-org-glm-4.7\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"zai-org-glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-flash-preview": {
"id": "gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-19",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.7,
"output": 3.75,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/gemini-3-flash-preview\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-instruct-2507": {
"id": "qwen3-235b-a22b-instruct-2507",
"name": "Qwen 3 235B A22B Instruct 2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-29",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen3-235b-a22b-instruct-2507\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-2603": {
"id": "mistral-small-2603",
"name": "Mistral Small 4",
"description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents",
"family": "mistral-small",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2026-03-16",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.1875,
"output": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/mistral-small-2603\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small-2603\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai-glm-5v-turbo": {
"id": "z-ai-glm-5v-turbo",
"name": "GLM 5V Turbo",
"description": "Fast GLM vision model for screenshots, documents, and multimodal agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32768
},
"cost": {
"input": 1.5,
"output": 5,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/z-ai-glm-5v-turbo\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"z-ai-glm-5v-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-6": {
"id": "grok-4-6",
"name": "Grok 4.6",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-10",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 200000
},
"cost": {
"input": 2.27,
"output": 6.8,
"cache_read": 0.57,
"tiers": [
{
"input": 4.53,
"output": 13.6,
"cache_read": 1.13,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4.53,
"output": 13.6,
"cache_read": 1.13
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/grok-4-6\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 6,
"output": 30,
"cache_read": 0.6,
"cache_write": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-opus-4-8\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-3-7-plus": {
"id": "qwen-3-7-plus",
"name": "Qwen 3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 2,
"cache_read": 0.05,
"cache_write": 0.625,
"tiers": [
{
"input": 1.5,
"output": 6,
"cache_read": 0.15,
"cache_write": 1.875,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 1.5,
"output": 6,
"cache_read": 0.15,
"cache_write": 1.875
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen-3-7-plus\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen-3-7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 1.65,
"output": 3.301,
"cache_read": 0.33
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/deepseek-v4-pro\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia-nemotron-3-ultra-550b-a55b": {
"id": "nvidia-nemotron-3-ultra-550b-a55b",
"name": "NVIDIA Nemotron 3 Ultra",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 32768
},
"cost": {
"input": 0.625,
"output": 3.125,
"cache_read": 0.1875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/nvidia-nemotron-3-ultra-550b-a55b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"nvidia-nemotron-3-ultra-550b-a55b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai-glm-5-3": {
"id": "z-ai-glm-5-3",
"name": "GLM 5.3",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-18",
"last_updated": "2026-08-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.75,
"output": 5.5,
"cache_read": 0.325
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/z-ai-glm-5-3\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"z-ai-glm-5-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-20-multi-agent": {
"id": "grok-4-20-multi-agent",
"name": "Grok 4.20 Multi-Agent",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": true,
"release_date": "2026-03-12",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 128000
},
"cost": {
"input": 1.42,
"output": 2.83,
"cache_read": 0.23,
"tiers": [
{
"input": 2.83,
"output": 5.67,
"cache_read": 0.45,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.83,
"output": 5.67,
"cache_read": 0.45
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/grok-4-20-multi-agent\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-20-multi-agent\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-56-luna": {
"id": "openai-gpt-56-luna",
"name": "GPT-5.6 Luna",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"cache_write": 0.3125,
"tiers": [
{
"input": 0.5,
"output": 2.25,
"cache_read": 0.05,
"cache_write": 0.625,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 0.5,
"output": 2.25,
"cache_read": 0.05,
"cache_write": 0.625
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-56-luna\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-56-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-70b": {
"id": "llama-3.3-70b",
"name": "Llama 3.3 70B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"release_date": "2025-04-06",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.7,
"output": 2.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/llama-3.3-70b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.3-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5": {
"id": "claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-29",
"last_updated": "2026-07-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/claude-sonnet-5\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-235b-a22b": {
"id": "qwen3-vl-235b-a22b",
"name": "Qwen3 VL 235B",
"description": "Multimodal model for analyzing text, images, documents, and rich media",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2026-01-16",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.21,
"output": 1.9,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/qwen3-vl-235b-a22b\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-235b-a22b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-4o-mini-2024-07-18": {
"id": "openai-gpt-4o-mini-2024-07-18",
"name": "GPT-4o Mini",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2026-02-28",
"last_updated": "2026-06-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.1875,
"output": 0.75,
"cache_read": 0.09375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-4o-mini-2024-07-18\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-4o-mini-2024-07-18\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-8-flash": {
"id": "gemini-3-8-flash",
"name": "Gemini 3.8 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.9375,
"output": 4.6875,
"cache_read": 0.09375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/gemini-3-8-flash\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai-gpt-56-sol-pro": {
"id": "openai-gpt-56-sol-pro",
"name": "GPT-5.6 Sol Pro",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 12.5,
"cache_read": 0.25,
"cache_write": 3.125,
"tiers": [
{
"input": 5,
"output": 18.75,
"cache_read": 0.5,
"cache_write": 6.25,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 18.75,
"cache_read": 0.5,
"cache_write": 6.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"venice/openai-gpt-56-sol-pro\", apiKey: processEnvironment[\"VENICE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"VENICE_API_KEY\"]\n)\nlet session = provider.model(\"openai-gpt-56-sol-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"gmicloud": {
"id": "gmicloud",
"name": "GMI Cloud",
"baseURL": "https://api.gmi-serving.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"GMICLOUD_API_KEY"
],
"doc": "https://docs.gmicloud.ai/inference-engine/api-reference/llm-api-reference",
"modelCount": 15,
"models": {
"deepseek-ai/DeepSeek-V4-Flash": {
"id": "deepseek-ai/DeepSeek-V4-Flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048575,
"output": 384000
},
"cost": {
"input": 0.112,
"output": 0.224,
"cache_read": 0.022
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/deepseek-ai/DeepSeek-V4-Flash\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro": {
"id": "deepseek-ai/DeepSeek-V4-Pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 1.392,
"output": 2.784,
"cache_read": 0.116
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/deepseek-ai/DeepSeek-V4-Pro\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.8": {
"id": "anthropic/claude-opus-4.8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/anthropic/claude-opus-4.8\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.7": {
"id": "anthropic/claude-opus-4.7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 409600,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 30,
"output": 150,
"cache_read": 3,
"cache_write": 37.5
},
"provider": {
"body": {
"speed": "fast"
},
"headers": {
"anthropic-beta": "fast-mode-2026-02-01"
}
}
}
}
},
"cost": {
"input": 4.5,
"output": 22.5,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/anthropic/claude-opus-4.7\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.6": {
"id": "anthropic/claude-sonnet-4.6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 409600,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/anthropic/claude-sonnet-4.6\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.6": {
"id": "anthropic/claude-opus-4.6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 127999
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 409600,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/anthropic/claude-opus-4.6\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2-FP8": {
"id": "zai-org/GLM-5.2-FP8",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.979,
"output": 3.08,
"cache_read": 0.182
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/zai-org/GLM-5.2-FP8\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5-FP8": {
"id": "zai-org/GLM-5-FP8",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 1.92,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/zai-org/GLM-5-FP8\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.1-FP8": {
"id": "zai-org/GLM-5.1-FP8",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 0.98,
"output": 3.08,
"cache_read": 0.182
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/zai-org/GLM-5.1-FP8\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.1-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.7-Max": {
"id": "Qwen/Qwen3.7-Max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.25,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/Qwen/Qwen3.7-Max\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.7-Max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M3": {
"id": "MiniMaxAI/MiniMax-M3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 512000
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.12,
"tiers": [
{
"input": 1.2,
"output": 4.8,
"cache_read": 0.24,
"tier": {
"type": "context",
"size": 512000
}
}
],
"context_over_200k": {
"input": 1.2,
"output": 4.8,
"cache_read": 0.24
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/MiniMaxAI/MiniMax-M3\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M2.7": {
"id": "MiniMaxAI/MiniMax-M2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/MiniMaxAI/MiniMax-M2.7\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/openai/gpt-5.5\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.7-code-highspeed": {
"id": "moonshotai/kimi-k2.7-code-highspeed",
"name": "Kimi K2.7 Code Highspeed",
"description": "Lower-latency Kimi Code variant for interactive edits and coding-agent loops",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.9,
"output": 8,
"cache_read": 0.38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/moonshotai/kimi-k2.7-code-highspeed\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.7-code-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6": {
"id": "moonshotai/Kimi-K2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.855,
"output": 3.6,
"cache_read": 0.144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"gmicloud/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"GMICLOUD_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.gmi-serving.com/v1\")!,\n apiKey: processEnvironment[\"GMICLOUD_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"io-net": {
"id": "io-net",
"name": "IO.NET",
"baseURL": "https://api.intelligence.io.solutions/api/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"IOINTELLIGENCE_API_KEY"
],
"doc": "https://io.net/docs/guides/intelligence/io-intelligence",
"modelCount": 17,
"models": {
"Intel/Qwen3-Coder-480B-A35B-Instruct-int4-mixed-ar": {
"id": "Intel/Qwen3-Coder-480B-A35B-Instruct-int4-mixed-ar",
"name": "Qwen 3 Coder 480B",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-01-15",
"last_updated": "2025-01-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 106000,
"output": 4096
},
"cost": {
"input": 0.22,
"output": 0.95,
"cache_read": 0.11,
"cache_write": 0.44
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/Intel/Qwen3-Coder-480B-A35B-Instruct-int4-mixed-ar\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"Intel/Qwen3-Coder-480B-A35B-Instruct-int4-mixed-ar\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-R1-0528": {
"id": "deepseek-ai/DeepSeek-R1-0528",
"name": "DeepSeek R1",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-01-20",
"last_updated": "2025-05-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 2,
"output": 8.75,
"cache_read": 1,
"cache_write": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/deepseek-ai/DeepSeek-R1-0528\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-R1-0528\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Devstral-Small-2505": {
"id": "mistralai/Devstral-Small-2505",
"name": "Devstral Small 2505",
"description": "Mistral coding agent model for repository tasks and software engineering workflows",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-05-01",
"last_updated": "2025-05-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.05,
"output": 0.22,
"cache_read": 0.025,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/mistralai/Devstral-Small-2505\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Devstral-Small-2505\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Mistral-Large-Instruct-2411": {
"id": "mistralai/Mistral-Large-Instruct-2411",
"name": "Mistral Large Instruct 2411",
"description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
"family": "mistral-large",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-11-01",
"last_updated": "2024-11-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 1,
"cache_write": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/mistralai/Mistral-Large-Instruct-2411\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Mistral-Large-Instruct-2411\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Mistral-Nemo-Instruct-2407": {
"id": "mistralai/Mistral-Nemo-Instruct-2407",
"name": "Mistral Nemo Instruct 2407",
"description": "Mistral model for multilingual chat, reasoning, and tool-assisted workflows",
"family": "mistral-nemo",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-05",
"release_date": "2024-07-01",
"last_updated": "2024-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.02,
"output": 0.04,
"cache_read": 0.01,
"cache_write": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/mistralai/Mistral-Nemo-Instruct-2407\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Mistral-Nemo-Instruct-2407\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Magistral-Small-2506": {
"id": "mistralai/Magistral-Small-2506",
"name": "Magistral Small 2506",
"description": "Mistral reasoning model for transparent analysis, math, and complex decisions",
"family": "magistral-small",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-01",
"last_updated": "2025-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 1.5,
"cache_read": 0.25,
"cache_write": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/mistralai/Magistral-Small-2506\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Magistral-Small-2506\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.6": {
"id": "zai-org/GLM-4.6",
"name": "GLM 4.6",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-11-15",
"last_updated": "2024-11-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 4096
},
"cost": {
"input": 0.4,
"output": 1.75,
"cache_read": 0.2,
"cache_write": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/zai-org/GLM-4.6\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen2.5-VL-32B-Instruct": {
"id": "Qwen/Qwen2.5-VL-32B-Instruct",
"name": "Qwen 2.5 VL 32B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-09",
"release_date": "2024-11-01",
"last_updated": "2024-11-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 4096
},
"cost": {
"input": 0.05,
"output": 0.22,
"cache_read": 0.025,
"cache_write": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/Qwen/Qwen2.5-VL-32B-Instruct\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen2.5-VL-32B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Next-80B-A3B-Instruct": {
"id": "Qwen/Qwen3-Next-80B-A3B-Instruct",
"name": "Qwen 3 Next 80B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-01-10",
"last_updated": "2025-01-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 4096
},
"cost": {
"input": 0.1,
"output": 0.8,
"cache_read": 0.05,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/Qwen/Qwen3-Next-80B-A3B-Instruct\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Next-80B-A3B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B-Thinking-2507": {
"id": "Qwen/Qwen3-235B-A22B-Thinking-2507",
"name": "Qwen 3 235B Thinking",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-07-01",
"last_updated": "2025-07-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 4096
},
"cost": {
"input": 0.11,
"output": 0.6,
"cache_read": 0.055,
"cache_write": 0.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/Qwen/Qwen3-235B-A22B-Thinking-2507\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B-Thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8": {
"id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
"name": "Llama 4 Maverick 17B 128E Instruct",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-01-15",
"last_updated": "2025-01-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 430000,
"output": 4096
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075,
"cache_write": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.2-90B-Vision-Instruct": {
"id": "meta-llama/Llama-3.2-90B-Vision-Instruct",
"name": "Llama 3.2 90B Vision Instruct",
"description": "Open Llama multimodal model for image understanding and text reasoning",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-09-25",
"last_updated": "2024-09-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 16000,
"output": 4096
},
"cost": {
"input": 0.35,
"output": 0.4,
"cache_read": 0.175,
"cache_write": 0.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/meta-llama/Llama-3.2-90B-Vision-Instruct\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.2-90B-Vision-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.3-70B-Instruct": {
"id": "meta-llama/Llama-3.3-70B-Instruct",
"name": "Llama 3.3 70B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.13,
"output": 0.38,
"cache_read": 0.065,
"cache_write": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/meta-llama/Llama-3.3-70B-Instruct\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.3-70B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "GPT-OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-12-01",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 64000,
"output": 4096
},
"cost": {
"input": 0.03,
"output": 0.14,
"cache_read": 0.015,
"cache_write": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/openai/gpt-oss-20b\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT-OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2024-12-01",
"last_updated": "2024-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 4096
},
"cost": {
"input": 0.04,
"output": 0.4,
"cache_read": 0.02,
"cache_write": 0.08
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/openai/gpt-oss-120b\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2-Thinking": {
"id": "moonshotai/Kimi-K2-Thinking",
"name": "Kimi K2 Thinking",
"description": "Kimi reasoning model for long-horizon research, planning, and tool use",
"family": "kimi-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2024-11-01",
"last_updated": "2024-11-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 4096
},
"cost": {
"input": 0.55,
"output": 2.25,
"cache_read": 0.275,
"cache_write": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/moonshotai/Kimi-K2-Thinking\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2-Thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2-Instruct-0905": {
"id": "moonshotai/Kimi-K2-Instruct-0905",
"name": "Kimi K2 Instruct",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2024-09-05",
"last_updated": "2024-09-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 4096
},
"cost": {
"input": 0.39,
"output": 1.9,
"cache_read": 0.195,
"cache_write": 0.78
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"io-net/moonshotai/Kimi-K2-Instruct-0905\", apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.intelligence.io.solutions/api/v1\")!,\n apiKey: processEnvironment[\"IOINTELLIGENCE_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2-Instruct-0905\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"llmgateway": {
"id": "llmgateway",
"name": "DevPass (LLM Gateway)",
"baseURL": "https://api.llmgateway.io/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"LLMGATEWAY_API_KEY"
],
"doc": "https://llmgateway.io/docs",
"modelCount": 188,
"models": {
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-sonnet-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ministral-14b-2512": {
"id": "ministral-14b-2512",
"name": "Ministral 14B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "mistral",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 8192
},
"cost": {
"input": 0.2,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/ministral-14b-2512\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"ministral-14b-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-nano": {
"id": "gpt-5-nano",
"name": "GPT-5 Nano",
"description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5-nano\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.7": {
"id": "glm-4.7",
"name": "GLM-4.7",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.38,
"output": 1.98,
"cache_read": 0.19,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-max": {
"id": "qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 3.75,
"cache_read": 0.25,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3.7-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b-it": {
"id": "gemma-4-26b-a4b-it",
"name": "Gemma 4 26B A4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.07,
"output": 0.34
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemma-4-26b-a4b-it\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.1-lightning": {
"id": "minimax-m2.1-lightning",
"name": "MiniMax M2.1 Lightning",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 131072
},
"cost": {
"input": 0.12,
"output": 0.48
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/minimax-m2.1-lightning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.1-lightning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-pro-latest": {
"id": "gemini-pro-latest",
"name": "Gemini Pro Latest",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-27",
"last_updated": "2026-02-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-pro-latest\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-pro-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code-highspeed": {
"id": "kimi-k2.7-code-highspeed",
"name": "Kimi K2.7 Code Highspeed",
"description": "Lower-latency Kimi Code variant for interactive edits and coding-agent loops",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 1.9,
"output": 8,
"cache_read": 0.38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/kimi-k2.7-code-highspeed\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.5-air": {
"id": "glm-4.5-air",
"name": "GLM-4.5-Air",
"description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
"family": "glm-air",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131000,
"output": 98304
},
"cost": {
"input": 0.13,
"output": 0.85,
"cache_read": 0.025,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4.5-air\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5-air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-1-fast-non-reasoning": {
"id": "grok-4-1-fast-non-reasoning",
"name": "Grok 4.1 Fast Non-Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 2000000
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/grok-4-1-fast-non-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-1-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"codestral-2508": {
"id": "codestral-2508",
"name": "Codestral",
"description": "Mistral coding model for code completion, generation, and developer workflows",
"family": "mistral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-30",
"last_updated": "2025-07-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0.3,
"output": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/codestral-2508\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"codestral-2508\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-nano": {
"id": "gpt-4.1-nano",
"name": "GPT-4.1 nano",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-4.1-nano\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-1-8-251228": {
"id": "seed-1-8-251228",
"name": "Seed 1.8 (251228)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-18",
"last_updated": "2025-12-18",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 8192
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/seed-1-8-251228\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"seed-1-8-251228\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1.3": {
"id": "muse-spark-1.3",
"name": "Muse Spark 1.3",
"description": "Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It improves long-horizon agent collaboration, instruction following, and coding efficiency relative to Muse Spark 1.2.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/muse-spark-1.3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-plus": {
"id": "qwen3-coder-plus",
"name": "Qwen3 Coder Plus",
"description": "Hosted Qwen coder for software agents, repo edits, and long-context code",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.2,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-coder-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-2506": {
"id": "mistral-small-2506",
"name": "Mistral Small 3.2",
"description": "Efficient Mistral model for fast chat, extraction, and production assistants",
"family": "mistral-small",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-03",
"release_date": "2025-06-20",
"last_updated": "2025-06-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.1,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/mistral-small-2506\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small-2506\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3.8-27B": {
"id": "Qwen3.8-27B",
"name": "Qwen3.8 27B",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/Qwen3.8-27B\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3.8-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-4-scout-17b-instruct": {
"id": "llama-4-scout-17b-instruct",
"name": "Llama 4 Scout 17B Instruct",
"description": "Open multimodal Llama model for long-context analysis and efficient agents",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 2048
},
"cost": {
"input": 0.18,
"output": 0.59
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/llama-4-scout-17b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"llama-4-scout-17b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen35-397b-a17b": {
"id": "qwen35-397b-a17b",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen35-397b-a17b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen35-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-next-80b-a3b-thinking": {
"id": "qwen3-next-80b-a3b-thinking",
"name": "Qwen3-Next 80B-A3B (Thinking)",
"description": "Efficient Qwen thinking model for local reasoning, math, and coding agents",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-next-80b-a3b-thinking\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-next-80b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1.2": {
"id": "muse-spark-1.2",
"name": "Muse Spark 1.2",
"description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/muse-spark-1.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-thinking-2507": {
"id": "qwen3-235b-a22b-thinking-2507",
"name": "Qwen3 235B A22B Thinking (2507)",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-08",
"last_updated": "2025-07-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 8192
},
"cost": {
"input": 0.3,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-235b-a22b-thinking-2507\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.6": {
"id": "glm-4.6",
"name": "GLM-4.6",
"description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.55,
"output": 2.2,
"cache_read": 0.11,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4.6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-pro": {
"id": "gpt-5-pro",
"name": "GPT-5 Pro",
"description": "Higher-accuracy GPT-5 tier for tough analysis, coding reviews, and planning",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 272000
},
"cost": {
"input": 15,
"output": 120
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o-mini-transcribe": {
"id": "gpt-4o-mini-transcribe",
"name": "GPT-4o Mini Transcribe",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-03-20",
"last_updated": "2025-03-20",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16000,
"output": 16000
},
"cost": {
"input": 1.25,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-4o-mini-transcribe\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4o-mini-transcribe\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.1": {
"id": "minimax-m2.1",
"name": "MiniMax-M2.1",
"description": "Earlier MiniMax agent model for practical coding and productivity tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.27,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/minimax-m2.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.6v": {
"id": "glm-4.6v",
"name": "GLM-4.6V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 0.9,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4.6v\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.6v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-9b": {
"id": "qwen3.5-9b",
"name": "Qwen3.5 9B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3.5-9b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-next-80b-a3b-instruct": {
"id": "qwen3-next-80b-a3b-instruct",
"name": "Qwen3-Next 80B-A3B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-flash": {
"id": "qwen3-coder-flash",
"name": "Qwen3 Coder Flash",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.5,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-coder-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.6v-flashx": {
"id": "glm-4.6v-flashx",
"name": "GLM-4.6V FlashX",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16000
},
"cost": {
"input": 0.04,
"output": 0.4,
"cache_read": 0.004
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4.6v-flashx\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.6v-flashx\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex-mini": {
"id": "gpt-5.1-codex-mini",
"name": "GPT-5.1 Codex mini",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.1-codex-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-codex-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-max": {
"id": "qwen-max",
"name": "Qwen Max",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-04-03",
"last_updated": "2025-01-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 1.6,
"output": 6.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-plus": {
"id": "qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"cache_write": 0.625,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"cache_write": 2.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3.6-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"devstral-2512": {
"id": "devstral-2512",
"name": "Devstral 2",
"description": "Mistral's coding-agent model for repository work, terminal tasks, and software fixes",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2025-12-09",
"last_updated": "2025-12-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"status": "deprecated",
"cost": {
"input": 0.4,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/devstral-2512\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"devstral-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ling-3.0-flash": {
"id": "ling-3.0-flash",
"name": "InclusionAI Ling 3.0 Flash",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "ling",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-02",
"last_updated": "2026-08-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.06,
"output": 0.18,
"cache_read": 0.012
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/ling-3.0-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"ling-3.0-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2": {
"id": "minimax-m2",
"name": "MiniMax-M2",
"description": "Efficient open MiniMax model built for coding agents and tool-heavy workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-10-27",
"last_updated": "2025-10-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 131072
},
"cost": {
"input": 0.2,
"output": 1,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/minimax-m2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1-codex": {
"id": "gpt-5.1-codex",
"name": "GPT-5.1 Codex",
"description": "Codex GPT for repository edits, code review, and practical software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.1-codex\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.7-highspeed": {
"id": "minimax-m2.7-highspeed",
"name": "MiniMax-M2.7-highspeed",
"description": "Low-latency M2.7 variant for interactive coding plans and agent loops",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.06,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/minimax-m2.7-highspeed\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-sol": {
"id": "gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.6-sol\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fugu-ultra": {
"id": "fugu-ultra",
"name": "Fugu Ultra",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-06-22",
"last_updated": "2026-06-22",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/fugu-ultra\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fugu-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1.2-contributor": {
"id": "muse-spark-1.2-contributor",
"name": "Muse Spark 1.2 Contributor",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-06",
"last_updated": "2026-08-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.002
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/muse-spark-1.2-contributor\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1.2-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-fp8": {
"id": "qwen3-235b-a22b-fp8",
"name": "Qwen3 235B A22B FP8",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-28",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 8192
},
"cost": {
"input": 0.2,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-235b-a22b-fp8\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5": {
"id": "claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-opus-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.7": {
"id": "minimax-m2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.08,
"output": 0.32,
"cache_read": 0.017,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/minimax-m2.7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"custom": {
"id": "custom",
"name": "Custom Model",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"family": "auto",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/custom\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"custom\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 3.05,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/kimi-k2.6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-flash": {
"id": "qwen-flash",
"name": "Qwen Flash",
"description": "Efficient Qwen model for fast chat, extraction, and high-volume workloads",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.01,
"cache_write": 0.0625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-pro-preview": {
"id": "gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2-codex": {
"id": "gpt-5.2-codex",
"name": "GPT-5.2 Codex",
"description": "Code-specialist GPT for repository edits, reviews, and long-running software agents",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.2-codex\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.2-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.8,
"output": 2.55,
"cache_read": 0.16,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-6-astra": {
"id": "gpt-6-astra",
"name": "GPT-6 Astra",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-09-03",
"last_updated": "2026-09-03",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"output": 1050000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-6-astra\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-6-astra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash-lite": {
"id": "gemini-2.5-flash-lite",
"name": "Gemini 2.5 Flash-Lite",
"description": "Lean Gemini 2.5 lane for cheap multimodal traffic and quick agents",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-2.5-flash-lite\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.5": {
"id": "minimax-m2.5",
"name": "MiniMax-M2.5",
"description": "Prior MiniMax coding model for agent workflows, office edits, and automation",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 228700,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/minimax-m2.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m3": {
"id": "minimax-m3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 512000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/minimax-m3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4": {
"id": "grok-4",
"name": "Grok 4",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-09",
"last_updated": "2025-07-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/grok-4\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1050000,
"output": 384000
},
"cost": {
"input": 0.05,
"output": 0.1,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/deepseek-v4-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code": {
"id": "kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/kimi-k2.7-code\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-1-6-flash-250715": {
"id": "seed-1-6-flash-250715",
"name": "Seed 1.6 Flash (250715)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-26",
"last_updated": "2025-07-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 8192
},
"cost": {
"input": 0.07,
"output": 0.3,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/seed-1-6-flash-250715\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"seed-1-6-flash-250715\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-thinking": {
"id": "kimi-k2-thinking",
"name": "Kimi K2 Thinking",
"description": "Thinking Kimi model for slower research passes, planning, and hard technical questions",
"family": "kimi-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-11-06",
"last_updated": "2025-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.6,
"output": 2.5,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/kimi-k2-thinking\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-32b": {
"id": "qwen3-32b",
"name": "Qwen3 32B",
"description": "Dense open Qwen model for self-hosted chat, reasoning, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 16384
},
"cost": {
"input": 0.36,
"output": 0.87,
"reasoning": 8.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-32b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.2-11b-instruct": {
"id": "llama-3.2-11b-instruct",
"name": "Llama 3.2 11B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2024-09-25",
"last_updated": "2024-09-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.07,
"output": 0.33
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/llama-3.2-11b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.2-11b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2-pro": {
"id": "gpt-5.2-pro",
"name": "GPT-5.2 Pro",
"description": "Higher-accuracy GPT-5.2 variant for tougher reasoning and review workflows",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 21,
"output": 168
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.2-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.2-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-1-20250805": {
"id": "claude-opus-4-1-20250805",
"name": "Claude Opus 4.1",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 31999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-opus-4-1-20250805\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-1-20250805\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2.5-highspeed": {
"id": "minimax-m2.5-highspeed",
"name": "MiniMax-M2.5-highspeed",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-13",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.03,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/minimax-m2.5-highspeed\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2.5-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.2-3b-instruct": {
"id": "llama-3.2-3b-instruct",
"name": "Llama 3.2 3B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2024-09-18",
"last_updated": "2024-09-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 32000
},
"cost": {
"input": 0.03,
"output": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/llama-3.2-3b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.2-3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1-mini": {
"id": "gpt-4.1-mini",
"name": "GPT-4.1 mini",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-4.1-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.6-flash": {
"id": "gemini-3.6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-3.6-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.5-x": {
"id": "glm-4.5-x",
"name": "GLM-4.5 X",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"status": "beta",
"cost": {
"input": 2.2,
"output": 8.9,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4.5-x\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5-x\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.4\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-20b": {
"id": "gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32766
},
"cost": {
"input": 0.04,
"output": 0.19,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-oss-20b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.1-flash-lite": {
"id": "gemini-3.1-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-07",
"last_updated": "2026-05-07",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-3.1-flash-lite\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o-transcribe": {
"id": "gpt-4o-transcribe",
"name": "GPT-4o Transcribe",
"description": "Speech transcription model for accurate audio-to-text and captioning workflows",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-03-20",
"last_updated": "2025-03-20",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16000,
"output": 16000
},
"cost": {
"input": 2.5,
"output": 10
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-4o-transcribe\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4o-transcribe\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4-turbo": {
"id": "gpt-4-turbo",
"name": "GPT-4 Turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 10,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-4-turbo\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-plus": {
"id": "qwen3-vl-plus",
"name": "Qwen3-VL Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 1.6,
"reasoning": 4.8,
"cache_read": 0.04,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-vl-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-1-fast-reasoning": {
"id": "grok-4-1-fast-reasoning",
"name": "Grok 4.1 Fast Reasoning",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/grok-4-1-fast-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-1-fast-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4.1-flash": {
"id": "deepseek-v4.1-flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1050000,
"output": 384000
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/deepseek-v4.1-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4.1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"hy3": {
"id": "hy3",
"name": "Hy3",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 192000,
"output": 128000
},
"cost": {
"input": 0.14,
"output": 0.58,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/hy3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-next": {
"id": "qwen3-coder-next",
"name": "Qwen3 Coder Next",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.108,
"output": 0.675,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-coder-next\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-coder-plus": {
"id": "qwen-coder-plus",
"name": "Qwen Coder Plus",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-09-18",
"last_updated": "2024-09-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.502,
"output": 1.004
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen-coder-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5-1": {
"id": "claude-fable-5-1",
"name": "Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-fable-5-1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-30b-a3b-instruct": {
"id": "qwen3-coder-30b-a3b-instruct",
"name": "Qwen3-Coder 30B-A3B Instruct",
"description": "Smaller Qwen coder for efficient local agents and repo-level fixes",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 65536
},
"cost": {
"input": 0.07,
"output": 0.27
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.1": {
"id": "gpt-5.1",
"name": "GPT-5.1",
"description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ernie-4.5-vl-424b-a47b": {
"id": "ernie-4.5-vl-424b-a47b",
"name": "ERNIE 4.5 VL 424B A47B",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "ernie",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-06-30",
"last_updated": "2025-06-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 123000,
"output": 123000
},
"cost": {
"input": 0.42,
"output": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/ernie-4.5-vl-424b-a47b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"ernie-4.5-vl-424b-a47b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-opus-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-large-2512": {
"id": "mistral-large-2512",
"name": "Mistral Large 3",
"description": "Mistral's largest general model for enterprise agents, coding, and multilingual reasoning",
"family": "mistral-large",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-11",
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.5,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/mistral-large-2512\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral-large-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash": {
"id": "gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-3.5-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-text-01": {
"id": "minimax-text-01",
"name": "MiniMax Text 01",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-01-15",
"last_updated": "2025-01-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.2,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/minimax-text-01\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"minimax-text-01\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o1": {
"id": "o1",
"name": "o1",
"description": "O-series reasoning model for hard analysis, math, coding, and planning",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2023-09",
"release_date": "2024-12-05",
"last_updated": "2024-12-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 15,
"output": 60,
"cache_read": 7.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/o1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"o1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.5-airx": {
"id": "glm-4.5-airx",
"name": "GLM-4.5 AirX",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 1.1,
"output": 4.5,
"cache_read": 0.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4.5-airx\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5-airx\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o": {
"id": "gpt-4o",
"name": "GPT-4o",
"description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 2.5,
"output": 10,
"cache_read": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-4o\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-luna": {
"id": "gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.6-luna\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ministral-3b-2512": {
"id": "ministral-3b-2512",
"name": "Ministral 3B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "mistral",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.1,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/ministral-3b-2512\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"ministral-3b-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5-20250929": {
"id": "claude-sonnet-4-5-20250929",
"name": "Claude Sonnet 4.5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-sonnet-4-5-20250929\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5-20250929\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-flash": {
"id": "qwen3.7-flash",
"name": "Qwen3.7 Flash",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-27",
"last_updated": "2026-07-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 0.03,
"output": 0.13,
"cache_read": 0.006,
"cache_write": 0.0375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3.7-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-7": {
"id": "claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-opus-4-7\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/kimi-k3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3-fast": {
"id": "kimi-k3-fast",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1040384,
"output": 131072
},
"cost": {
"input": 4.5,
"output": 22.5,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/kimi-k3-fast\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3.2": {
"id": "deepseek-v3.2",
"name": "DeepSeek V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 16384
},
"cost": {
"input": 0.26,
"output": 0.38,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/deepseek-v3.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-35b-a3b": {
"id": "qwen3.6-35b-a3b",
"name": "Qwen3.6 35B-A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.248,
"output": 1.485
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3.6-35b-a3b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nemotron-3-ultra-550b": {
"id": "nemotron-3-ultra-550b",
"name": "Nemotron 3 Ultra 550B A55B",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 128000
},
"cost": {
"input": 0.5,
"output": 2.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/nemotron-3-ultra-550b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"nemotron-3-ultra-550b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2-fast": {
"id": "glm-5.2-fast",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2.2,
"output": 6.5,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-5.2-fast\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-max": {
"id": "qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.845,
"output": 3.38,
"cache_read": 0.6,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.3-codex": {
"id": "gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.3-codex\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-5": {
"id": "grok-4-5",
"name": "Grok 4.5",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/grok-4-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5-20251001": {
"id": "claude-haiku-4-5-20251001",
"name": "Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-haiku-4-5-20251001\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5-20251001\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.088,
"output": 0.25,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-5.3-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4o-mini": {
"id": "gpt-4o-mini",
"name": "GPT-4o mini",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-4o-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.5": {
"id": "glm-4.5",
"name": "GLM-4.5",
"description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131000,
"output": 98304
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1.3-contributor": {
"id": "muse-spark-1.3-contributor",
"name": "Muse Spark 1.3 Contributor",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.002
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/muse-spark-1.3-contributor\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1.3-contributor\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-fable-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash-lite": {
"id": "gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-flash": {
"id": "qwen3-vl-flash",
"name": "Qwen3 VL Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-09",
"last_updated": "2025-10-09",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-vl-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-30b-a3b-instruct": {
"id": "qwen3-vl-30b-a3b-instruct",
"name": "Qwen3 VL 30B A3B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-02",
"last_updated": "2025-10-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 8192
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-vl-30b-a3b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-plus": {
"id": "qwen-plus",
"name": "Qwen Plus",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01-25",
"last_updated": "2025-09-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 1.2,
"reasoning": 4,
"cache_read": 0.08,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4.1": {
"id": "gpt-4.1",
"name": "GPT-4.1",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-4.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar": {
"id": "sonar",
"name": "Sonar",
"description": "Fast web-grounded Sonar for current answers, citations, and lightweight retrieval",
"family": "sonar",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 130000,
"output": 4096
},
"cost": {
"input": 1,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/sonar\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"sonar\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-nano": {
"id": "gpt-5.4-nano",
"name": "GPT-5.4 nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.4-nano\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5-pro": {
"id": "gpt-5.5-pro",
"name": "GPT-5.5 Pro",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180,
"tiers": [
{
"input": 60,
"output": 270,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 60,
"output": 270
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.5-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar-reasoning-pro": {
"id": "sonar-reasoning-pro",
"name": "Sonar Reasoning Pro",
"description": "Web-grounded Sonar for multi-step research questions that need cited reasoning",
"family": "sonar-reasoning",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 2,
"output": 8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/sonar-reasoning-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"sonar-reasoning-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-4-maverick-17b-instruct": {
"id": "llama-4-maverick-17b-instruct",
"name": "Llama 4 Maverick 17B Instruct",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 2048
},
"cost": {
"input": 0.27,
"output": 0.85
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/llama-4-maverick-17b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"llama-4-maverick-17b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1.1": {
"id": "muse-spark-1.1",
"name": "Muse Spark 1.1",
"description": "Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-08",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/muse-spark-1.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.5v": {
"id": "glm-4.5v",
"name": "GLM-4.5V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-08-11",
"last_updated": "2025-08-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.6,
"output": 1.8,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4.5v\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-large-latest": {
"id": "mistral-large-latest",
"name": "Mistral Large (latest)",
"description": "Flagship Mistral model for advanced reasoning, coding, and multilingual work",
"family": "mistral-large",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-11",
"release_date": "2024-11-01",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 262144
},
"cost": {
"input": 4,
"output": 12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/mistral-large-latest\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mistral-large-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-flash": {
"id": "qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.05,
"cache_write": 0.3125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3.6-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-build-0-1": {
"id": "grok-build-0-1",
"name": "Grok Build 0.1",
"description": "Fast Grok coding model tuned for agentic engineering and iterative edits",
"family": "grok-build",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 1,
"output": 2,
"cache_read": 0.2,
"tiers": [
{
"input": 2,
"output": 4,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2,
"output": 4,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/grok-build-0-1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"grok-build-0-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.7-flashx": {
"id": "glm-4.7-flashx",
"name": "GLM-4.7-FlashX",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0.07,
"output": 0.4,
"cache_read": 0.01,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4.7-flashx\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.7-flashx\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-flash": {
"id": "qwen3.8-flash",
"name": "Qwen3.8 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.47,
"cache_read": 0.016,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3.8-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fugu-max": {
"id": "fugu-max",
"name": "Fugu Max",
"description": "Multi-agent model for routing expert agents across complex analytical tasks",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-11",
"last_updated": "2026-09-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/fugu-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fugu-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-mini": {
"id": "gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.4-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-3": {
"id": "grok-4-3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/grok-4-3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-max-preview": {
"id": "qwen3.6-max-preview",
"name": "Qwen3.6 Max Preview",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-20",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 1.3,
"output": 7.8,
"cache_read": 0.13,
"cache_write": 1.625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3.6-max-preview\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2": {
"id": "kimi-k2",
"name": "Kimi K2",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-11",
"last_updated": "2025-07-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0.57,
"output": 2.3,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/kimi-k2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it": {
"id": "gemma-4-31b-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.25,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemma-4-31b-it\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4-5": {
"id": "claude-haiku-4-5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-haiku-4-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4-5": {
"id": "claude-sonnet-4-5",
"name": "Claude Sonnet 4.5 (latest)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 63999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-sonnet-4-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5": {
"id": "glm-5",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 203000,
"output": 131072
},
"cost": {
"input": 0.72,
"output": 2.3,
"cache_read": 0.144,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4-32b-0414-128k": {
"id": "glm-4-32b-0414-128k",
"name": "GLM-4 32B (0414-128k)",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0.1,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4-32b-0414-128k\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4-32b-0414-128k\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-1-6-250615": {
"id": "seed-1-6-250615",
"name": "Seed 1.6 (250615)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-06-25",
"last_updated": "2025-06-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 8192
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/seed-1-6-250615\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"seed-1-6-250615\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-max": {
"id": "qwen3.8-max",
"name": "Qwen3.8 Max Preview",
"description": "Preview Qwen flagship for million-token multimodal reasoning and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-19",
"last_updated": "2026-07-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3.8-max\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.5": {
"id": "kimi-k2.5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.405,
"output": 1.98,
"cache_read": 0.225
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/kimi-k2.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-3.5-turbo": {
"id": "gpt-3.5-turbo",
"name": "GPT-3.5-turbo",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2021-09-01",
"release_date": "2023-03-01",
"last_updated": "2023-11-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 16385,
"output": 4096
},
"cost": {
"input": 0.5,
"output": 1.5,
"cache_read": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-3.5-turbo\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-3.5-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.931,
"output": 2.93,
"cache_read": 0.173,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-5.1\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-235b-a22b-thinking": {
"id": "qwen3-vl-235b-a22b-thinking",
"name": "Qwen3 VL 235B A22B Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-15",
"last_updated": "2025-09-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.98,
"output": 3.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-vl-235b-a22b-thinking\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-235b-a22b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-vl-235b-a22b-instruct": {
"id": "qwen3-vl-235b-a22b-instruct",
"name": "Qwen3 VL 235B A22B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-15",
"last_updated": "2025-09-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 8192
},
"cost": {
"input": 0.2,
"output": 0.88,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-vl-235b-a22b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-vl-235b-a22b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3-flash-preview": {
"id": "gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-3-flash-preview\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.1-70b-instruct": {
"id": "llama-3.1-70b-instruct",
"name": "Llama 3.1 70B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 2048
},
"status": "beta",
"cost": {
"input": 0.72,
"output": 0.72
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/llama-3.1-70b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.1-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.7-plus": {
"id": "qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.08,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3.7-plus\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-235b-a22b-instruct-2507": {
"id": "qwen3-235b-a22b-instruct-2507",
"name": "Qwen3 235B A22B Instruct (2507)",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-08",
"last_updated": "2025-07-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 8192
},
"cost": {
"input": 0.09,
"output": 0.58
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-235b-a22b-instruct-2507\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-235b-a22b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-omni-turbo": {
"id": "qwen-omni-turbo",
"name": "Qwen-Omni Turbo",
"description": "Qwen omni model for text, vision, audio, and multimodal agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-01-19",
"last_updated": "2025-03-26",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 2048
},
"cost": {
"input": 0.2,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen-omni-turbo\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen-omni-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-20-beta-0309-reasoning": {
"id": "grok-4-20-beta-0309-reasoning",
"name": "Grok 4.20 (Reasoning)",
"description": "Reasoning Grok for document-heavy analysis and long-horizon tool use",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/grok-4-20-beta-0309-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-20-beta-0309-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-6": {
"id": "grok-4-6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/grok-4-6\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.8-flash": {
"id": "gemini-3.8-flash",
"name": "Gemini 3.8 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-3.8-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-8": {
"id": "claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-opus-4-8\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-20-beta-0309-non-reasoning": {
"id": "grok-4-20-beta-0309-non-reasoning",
"name": "Grok 4.20 (Non-Reasoning)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/grok-4-20-beta-0309-non-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-20-beta-0309-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1050000,
"output": 384000
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.003625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/deepseek-v4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sonar-pro": {
"id": "sonar-pro",
"name": "Sonar Pro",
"description": "Deeper Sonar search model with broader retrieval and stronger synthesis",
"family": "sonar-pro",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 3,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/sonar-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"sonar-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ministral-8b-2512": {
"id": "ministral-8b-2512",
"name": "Ministral 8B",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "mistral",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 8192
},
"cost": {
"input": 0.15,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/ministral-8b-2512\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"ministral-8b-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-mini": {
"id": "gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32766
},
"cost": {
"input": 0.032,
"output": 0.14,
"cache_read": 0.032
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-oss-120b\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-pro": {
"id": "gpt-5.4-pro",
"name": "GPT-5.4 Pro",
"description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.4-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-coder-480b-a35b-instruct": {
"id": "qwen3-coder-480b-a35b-instruct",
"name": "Qwen3-Coder 480B-A35B Instruct",
"description": "Open Qwen coding heavyweight for repository reasoning and agentic engineering",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.38,
"output": 1.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen3-coder-480b-a35b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-coder-480b-a35b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-1-6-250915": {
"id": "seed-1-6-250915",
"name": "Seed 1.6 (250915)",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-15",
"last_updated": "2025-09-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 8192
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/seed-1-6-250915\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"seed-1-6-250915\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.7-flash": {
"id": "gemini-3.7-flash",
"name": "Gemini 3.7 Flash",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075,
"cache_write": 0.08333
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-3.7-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-pro": {
"id": "gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-2.5-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-terra": {
"id": "gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.6-terra\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-5.3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-4": {
"id": "gpt-4",
"name": "GPT-4",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2023-11",
"release_date": "2023-11-06",
"last_updated": "2024-04-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 8192,
"output": 8192
},
"cost": {
"input": 30,
"output": 60
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-4\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-20-non-reasoning": {
"id": "grok-4-20-non-reasoning",
"name": "Grok 4.20 (Non-Reasoning)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/grok-4-20-non-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-20-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.2": {
"id": "gpt-5.2",
"name": "GPT-5.2",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.2\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen-plus-latest": {
"id": "qwen-plus-latest",
"name": "Qwen Plus Latest",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-01-25",
"last_updated": "2025-01-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 8192
},
"cost": {
"input": 0.4,
"output": 1.2,
"cache_read": 0.08,
"cache_write": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/qwen-plus-latest\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"qwen-plus-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5": {
"id": "gpt-5",
"name": "GPT-5",
"description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-2.5-flash": {
"id": "gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1,
"max": 24576
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gemini-2.5-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fugu-ultra-v2.0": {
"id": "fugu-ultra-v2.0",
"name": "Fugu Ultra v2.0",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-11",
"last_updated": "2026-09-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/fugu-ultra-v2.0\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"fugu-ultra-v2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5": {
"id": "claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": false,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-sonnet-5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3.3-70b-instruct": {
"id": "llama-3.3-70b-instruct",
"name": "Llama-3.3-70B-Instruct",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 4096
},
"cost": {
"input": 0.135,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/llama-3.3-70b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"llama-3.3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"llama-3-70b-instruct": {
"id": "llama-3-70b-instruct",
"name": "Llama 3 70B Instruct",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2024-04-18",
"last_updated": "2024-04-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8000
},
"cost": {
"input": 0.51,
"output": 0.74
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/llama-3-70b-instruct\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"llama-3-70b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4-20-reasoning": {
"id": "grok-4-20-reasoning",
"name": "Grok 4.20 (Reasoning)",
"description": "Reasoning Grok for document-heavy analysis and long-horizon tool use",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/grok-4-20-reasoning\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"grok-4-20-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o4-mini": {
"id": "o4-mini",
"name": "o4-mini",
"description": "Fast o-series model for compact reasoning, coding, and tool use",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/o4-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"o4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5": {
"id": "mimo-v2.5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028,
"tiers": [
{
"input": 0.8,
"output": 4,
"cache_read": 0.16,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 0.8,
"output": 4,
"cache_read": 0.16
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/mimo-v2.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o3-mini": {
"id": "o3-mini",
"name": "o3-mini",
"description": "Smaller o-series reasoner for economical coding, math, and planning tasks",
"family": "o-mini",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2024-12-20",
"last_updated": "2025-01-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/o3-mini\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-pro": {
"id": "mimo-v2.5-pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.435,
"output": 0.87,
"cache_read": 0.0036,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/mimo-v2.5-pro\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.7-flash": {
"id": "glm-4.7-flash",
"name": "GLM-4.7-Flash",
"description": "Budget GLM lane for fast coding help, routing, and everyday automation",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0.06,
"output": 0.4,
"cache_read": 0.01,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/glm-4.7-flash\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4-5-20251101": {
"id": "claude-opus-4-5-20251101",
"name": "Claude Opus 4.5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 31999
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-01",
"last_updated": "2025-11-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/claude-opus-4-5-20251101\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"claude-opus-4-5-20251101\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"o3": {
"id": "o3",
"name": "o3",
"description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/o3\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"auto": {
"id": "auto",
"name": "Auto Route",
"description": "Automatic model router for matching prompts to suitable backends and budgets",
"family": "auto",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-01-01",
"last_updated": "2024-01-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/auto\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"auto\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"llmgateway/gpt-5.5\", apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.llmgateway.io/v1\")!,\n apiKey: processEnvironment[\"LLMGATEWAY_API_KEY\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"infomaniak": {
"id": "infomaniak",
"name": "Infomaniak",
"baseURL": "https://api.infomaniak.com/2/ai/${INFOMANIAK_PRODUCT_ID}/openai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"INFOMANIAK_API_KEY",
"INFOMANIAK_PRODUCT_ID"
],
"doc": "https://www.infomaniak.com/en/hosting/ai-services/open-source-models",
"modelCount": 10,
"models": {
"bge_multilingual_gemma2": {
"id": "bge_multilingual_gemma2",
"name": "BGE Multilingual Gemma2",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2024-07-25",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8000,
"input": 8000,
"output": 3584
},
"cost": {
"input": 0.08,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"infomaniak/bge_multilingual_gemma2\", apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.infomaniak.com/2/ai/${INFOMANIAK_PRODUCT_ID}/openai/v1\")!,\n apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"]\n)\nlet session = provider.model(\"bge_multilingual_gemma2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mini_lm_l12_v2": {
"id": "mini_lm_l12_v2",
"name": "All-MiniLM-L12-v2",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2021-08-30",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128,
"input": 128,
"output": 384
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"infomaniak/mini_lm_l12_v2\", apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.infomaniak.com/2/ai/${INFOMANIAK_PRODUCT_ID}/openai/v1\")!,\n apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"]\n)\nlet session = provider.model(\"mini_lm_l12_v2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"swiss-ai/Apertus-v1.5-70B": {
"id": "swiss-ai/Apertus-v1.5-70B",
"name": "Apertus v1.5 70B",
"description": "Open, ethically-sourced Swiss AI model for multilingual, multimodal chat and instruction following",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2026-07-24",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 100000,
"input": 100000,
"output": 8192
},
"status": "beta",
"cost": {
"input": 0.87,
"output": 3.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"infomaniak/swiss-ai/Apertus-v1.5-70B\", apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.infomaniak.com/2/ai/${INFOMANIAK_PRODUCT_ID}/openai/v1\")!,\n apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"]\n)\nlet session = provider.model(\"swiss-ai/Apertus-v1.5-70B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Mistral-Small-4-119B-2603": {
"id": "mistralai/Mistral-Small-4-119B-2603",
"name": "Mistral Small 4",
"description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents",
"family": "mistral-small",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2026-03-16",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"cost": {
"input": 0.25,
"output": 0.93
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"infomaniak/mistralai/Mistral-Small-4-119B-2603\", apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.infomaniak.com/2/ai/${INFOMANIAK_PRODUCT_ID}/openai/v1\")!,\n apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Mistral-Small-4-119B-2603\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistralai/Ministral-3-14B-Instruct-2512": {
"id": "mistralai/Ministral-3-14B-Instruct-2512",
"name": "Ministral 3 14B Instruct",
"description": "Compact Mistral model for edge, latency-sensitive, and cost-efficient workloads",
"family": "ministral",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-02",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 100000,
"input": 100000,
"output": 25600
},
"status": "beta",
"cost": {
"input": 0.37,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"infomaniak/mistralai/Ministral-3-14B-Instruct-2512\", apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.infomaniak.com/2/ai/${INFOMANIAK_PRODUCT_ID}/openai/v1\")!,\n apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"]\n)\nlet session = provider.model(\"mistralai/Ministral-3-14B-Instruct-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8": {
"id": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8",
"name": "Nemotron 3 Nano 30B A3B FP8",
"description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"temperature": true,
"release_date": "2025-12-15",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"input": 1000000,
"output": 262144
},
"status": "beta",
"cost": {
"input": 0.06,
"output": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"infomaniak/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8\", apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.infomaniak.com/2/ai/${INFOMANIAK_PRODUCT_ID}/openai/v1\")!,\n apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31B-it": {
"id": "google/gemma-4-31B-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 100000,
"input": 100000,
"output": 32768
},
"cost": {
"input": 0.25,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"infomaniak/google/gemma-4-31B-it\", apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.infomaniak.com/2/ai/${INFOMANIAK_PRODUCT_ID}/openai/v1\")!,\n apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-122B-A10B-FP8": {
"id": "Qwen/Qwen3.5-122B-A10B-FP8",
"name": "Qwen3.5 122B-A10B FP8",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3.97
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"infomaniak/Qwen/Qwen3.5-122B-A10B-FP8\", apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.infomaniak.com/2/ai/${INFOMANIAK_PRODUCT_ID}/openai/v1\")!,\n apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-122B-A10B-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-397B-A17B-FP8": {
"id": "Qwen/Qwen3.5-397B-A17B-FP8",
"name": "Qwen3.5 397B-A17B FP8",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"input": 200000,
"output": 65536
},
"status": "beta",
"cost": {
"input": 0.99,
"output": 4.46
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"infomaniak/Qwen/Qwen3.5-397B-A17B-FP8\", apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.infomaniak.com/2/ai/${INFOMANIAK_PRODUCT_ID}/openai/v1\")!,\n apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-397B-A17B-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6": {
"id": "moonshotai/Kimi-K2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 256000
},
"status": "beta",
"cost": {
"input": 0.74,
"output": 3.72
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"infomaniak/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.infomaniak.com/2/ai/${INFOMANIAK_PRODUCT_ID}/openai/v1\")!,\n apiKey: processEnvironment[\"INFOMANIAK_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"inception": {
"id": "inception",
"name": "Inception",
"baseURL": "https://api.inceptionlabs.ai/v1/",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"INCEPTION_API_KEY"
],
"doc": "https://platform.inceptionlabs.ai/docs",
"modelCount": 3,
"models": {
"mercury-2.5": {
"id": "mercury-2.5",
"name": "Mercury 2.5",
"description": "Mercury 2.5 is the fastest reasoning LLM, and the latest diffusion LLM (dLLM) from Inception",
"family": "mercury",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-11-01",
"release_date": "2026-09-08",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 260000,
"output": 65536
},
"cost": {
"input": 0.04,
"output": 0.15,
"cache_read": 0.004
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inception/mercury-2.5\", apiKey: processEnvironment[\"INCEPTION_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inceptionlabs.ai/v1/\")!,\n apiKey: processEnvironment[\"INCEPTION_API_KEY\"]\n)\nlet session = provider.model(\"mercury-2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mercury-edit-2": {
"id": "mercury-edit-2",
"name": "Mercury Edit 2",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"release_date": "2026-03-30",
"last_updated": "2026-03-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.25,
"output": 0.75,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inception/mercury-edit-2\", apiKey: processEnvironment[\"INCEPTION_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inceptionlabs.ai/v1/\")!,\n apiKey: processEnvironment[\"INCEPTION_API_KEY\"]\n)\nlet session = provider.model(\"mercury-edit-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mercury-2": {
"id": "mercury-2",
"name": "Mercury 2",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"family": "mercury",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01-01",
"release_date": "2026-02-24",
"last_updated": "2026-02-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 50000
},
"cost": {
"input": 0.25,
"output": 0.75,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inception/mercury-2\", apiKey: processEnvironment[\"INCEPTION_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inceptionlabs.ai/v1/\")!,\n apiKey: processEnvironment[\"INCEPTION_API_KEY\"]\n)\nlet session = provider.model(\"mercury-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"lilac": {
"id": "lilac",
"name": "Lilac",
"baseURL": "https://api.getlilac.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"LILAC_API_KEY"
],
"doc": "https://docs.getlilac.com/inference/models",
"modelCount": 4,
"models": {
"google/gemma-4-31b-it": {
"id": "google/gemma-4-31b-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262100,
"output": 262100
},
"cost": {
"input": 0.11,
"output": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"lilac/google/gemma-4-31b-it\", apiKey: processEnvironment[\"LILAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.getlilac.com/v1\")!,\n apiKey: processEnvironment[\"LILAC_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/glm-5.2": {
"id": "zai-org/glm-5.2",
"name": "GLM 5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 524288
},
"cost": {
"input": 0.9,
"output": 3,
"cache_read": 0.27
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"lilac/zai-org/glm-5.2\", apiKey: processEnvironment[\"LILAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.getlilac.com/v1\")!,\n apiKey: processEnvironment[\"LILAC_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimaxai/minimax-m3": {
"id": "minimaxai/minimax-m3",
"name": "MiniMax M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax-m3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 0.28,
"output": 1.1,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"lilac/minimaxai/minimax-m3\", apiKey: processEnvironment[\"LILAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.getlilac.com/v1\")!,\n apiKey: processEnvironment[\"LILAC_API_KEY\"]\n)\nlet session = provider.model(\"minimaxai/minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.6": {
"id": "moonshotai/kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.7,
"output": 3.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"lilac/moonshotai/kimi-k2.6\", apiKey: processEnvironment[\"LILAC_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.getlilac.com/v1\")!,\n apiKey: processEnvironment[\"LILAC_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"fastrouter": {
"id": "fastrouter",
"name": "FastRouter",
"baseURL": "https://go.fastrouter.ai/api/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"FASTROUTER_API_KEY"
],
"doc": "https://fastrouter.ai/models",
"modelCount": 47,
"models": {
"qwen/qwen3-coder": {
"id": "qwen/qwen3-coder",
"name": "Qwen3 Coder",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 66536
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/qwen/qwen3-coder\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/deepseek-r1-distill-llama-70b": {
"id": "deepseek-ai/deepseek-r1-distill-llama-70b",
"name": "DeepSeek R1 Distill Llama 70B",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-01-23",
"last_updated": "2025-01-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.03,
"output": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/deepseek-ai/deepseek-r1-distill-llama-70b\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/deepseek-r1-distill-llama-70b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7-highspeed": {
"id": "minimax/minimax-m2.7-highspeed",
"name": "MiniMax-M2.7-highspeed",
"description": "Low-latency M2.7 variant for interactive coding plans and agent loops",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/minimax/minimax-m2.7-highspeed\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax/minimax-m2.7": {
"id": "minimax/minimax-m2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/minimax/minimax-m2.7\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"minimax/minimax-m2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.8": {
"id": "anthropic/claude-opus-4.8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 32000
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/anthropic/claude-opus-4.8\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.1": {
"id": "anthropic/claude-opus-4.1",
"name": "Claude Opus 4.1",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 32000
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/anthropic/claude-opus-4.1\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.6": {
"id": "anthropic/claude-sonnet-4.6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 32000
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/anthropic/claude-sonnet-4.6\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4": {
"id": "anthropic/claude-sonnet-4",
"name": "Claude Sonnet 4",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 32000
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-05-22",
"last_updated": "2025-05-22",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/anthropic/claude-sonnet-4\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo3.1-fast": {
"id": "google/veo3.1-fast",
"name": "Veo 3.1 Fast",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-05-01",
"last_updated": "2026-05-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/veo3.1-fast\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/veo3.1-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo3.1": {
"id": "google/veo3.1",
"name": "Veo 3.1",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-05-01",
"last_updated": "2026-05-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/veo3.1\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/veo3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-pro-preview": {
"id": "google/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/veo3.1-lite": {
"id": "google/veo3.1-lite",
"name": "Veo 3.1 Lite",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "veo",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-05-01",
"last_updated": "2026-05-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/veo3.1-lite\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/veo3.1-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.5-flash": {
"id": "google/gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/gemini-3.5-flash\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/imagen-4.0-ultra": {
"id": "google/imagen-4.0-ultra",
"name": "Imagen 4 Ultra",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "imagen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-05-20",
"last_updated": "2025-05-20",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/imagen-4.0-ultra\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/imagen-4.0-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/imagen-4.0-fast": {
"id": "google/imagen-4.0-fast",
"name": "Imagen 4 Fast",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "imagen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-05-20",
"last_updated": "2025-05-20",
"modalities": {
"input": [
"text"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 480,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/imagen-4.0-fast\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/imagen-4.0-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3-pro-image-preview": {
"id": "google/gemini-3-pro-image-preview",
"name": "Nano Banana Pro Preview",
"description": "Nano Banana Pro for higher-fidelity image generation and design-heavy edits",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-11-20",
"last_updated": "2025-11-20",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 32768
},
"cost": {
"input": 2,
"output": 12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/gemini-3-pro-image-preview\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3-pro-image-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31b-it": {
"id": "google/gemma-4-31b-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.13,
"output": 0.38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/gemma-4-31b-it\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-pro": {
"id": "google/gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.31
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/gemini-2.5-pro\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-flash-image-preview": {
"id": "google/gemini-3.1-flash-image-preview",
"name": "Nano Banana 2 Preview",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-26",
"last_updated": "2026-02-26",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text",
"image"
]
},
"open_weights": false,
"limit": {
"context": 65536,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/gemini-3.1-flash-image-preview\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-flash-image-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash": {
"id": "google/gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.0375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/google/gemini-2.5-flash\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"bytedance/seedance-2": {
"id": "bytedance/seedance-2",
"name": "Seedance 2",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"family": "seed",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2026-04-01",
"last_updated": "2026-04-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": false,
"limit": {
"context": 4096,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/bytedance/seedance-2\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"bytedance/seedance-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"wanx/wan-v2-6": {
"id": "wanx/wan-v2-6",
"name": "Wan 2.6",
"description": "Video model for prompt-guided generation, editing, and motion workflows",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"video"
]
},
"open_weights": true,
"limit": {
"context": 400000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/wanx/wan-v2-6\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"wanx/wan-v2-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro": {
"id": "deepseek/deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 1.74,
"output": 3.48
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/deepseek/deepseek-v4-pro\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"leonardo-ai/lucid-realism": {
"id": "leonardo-ai/lucid-realism",
"name": "Lucid Realism",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "lucid",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-06-01",
"last_updated": "2025-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 4096,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/leonardo-ai/lucid-realism\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"leonardo-ai/lucid-realism\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"leonardo-ai/lucid-origin": {
"id": "leonardo-ai/lucid-origin",
"name": "Lucid Origin",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "lucid",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-06-01",
"last_updated": "2025-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 4096,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/leonardo-ai/lucid-origin\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"leonardo-ai/lucid-origin\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4.3": {
"id": "x-ai/grok-4.3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/x-ai/grok-4.3\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-4": {
"id": "x-ai/grok-4",
"name": "Grok 4",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-07-09",
"last_updated": "2025-07-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.75,
"cache_write": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/x-ai/grok-4\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"x-ai/grok-build-0.1": {
"id": "x-ai/grok-build-0.1",
"name": "Grok Build 0.1",
"description": "Fast Grok coding model tuned for agentic engineering and iterative edits",
"family": "grok-build",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 1,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/x-ai/grok-build-0.1\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"x-ai/grok-build-0.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-nano": {
"id": "openai/gpt-5-nano",
"name": "GPT-5 Nano",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10-01",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-5-nano\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-realtime-1.5": {
"id": "openai/gpt-realtime-1.5",
"name": "GPT Realtime 1.5",
"description": "Speech generation model for controllable voice, narration, and audio delivery",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-06-01",
"last_updated": "2025-06-01",
"modalities": {
"input": [
"text",
"audio",
"image"
],
"output": [
"text",
"audio"
]
},
"open_weights": false,
"limit": {
"context": 32000,
"output": 4096
},
"cost": {
"input": 4,
"output": 16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-realtime-1.5\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-realtime-1.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 65536
},
"cost": {
"input": 0.05,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-oss-20b\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-codex": {
"id": "openai/gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-5.3-codex\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1": {
"id": "openai/gpt-4.1",
"name": "GPT-4.1",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-4.1\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-nano": {
"id": "openai/gpt-5.4-nano",
"name": "GPT-5.4 nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-5.4-nano\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5-pro": {
"id": "openai/gpt-5.5-pro",
"name": "GPT-5.5 Pro",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-5.5-pro\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-mini": {
"id": "openai/gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-5.4-mini\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-image-2": {
"id": "openai/gpt-image-2",
"name": "GPT Image 2",
"description": "Image model for prompt-driven generation, editing, and visual design workflows",
"family": "gpt-image",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"image"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-image-2\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-image-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-mini": {
"id": "openai/gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10-01",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-5-mini\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-oss-120b\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5": {
"id": "openai/gpt-5",
"name": "GPT-5",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-10-01",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-5\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/openai/gpt-5.5\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2.6": {
"id": "moonshotai/kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.75,
"output": 3.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/moonshotai/kimi-k2.6\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k2": {
"id": "moonshotai/kimi-k2",
"name": "Kimi K2",
"description": "Kimi model for long-context chat, coding, and agentic reasoning",
"family": "kimi-k2",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-10",
"release_date": "2025-07-11",
"last_updated": "2025-07-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.55,
"output": 2.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/moonshotai/kimi-k2\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sarvam/sarvam-105b": {
"id": "sarvam/sarvam-105b",
"name": "Sarvam 105B",
"description": "Flagship Indian-language reasoning model for enterprise multilingual applications",
"family": "sarvam",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-09-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.04,
"output": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/sarvam/sarvam-105b\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"sarvam/sarvam-105b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sarvam/sarvam-30b": {
"id": "sarvam/sarvam-30b",
"name": "Sarvam 30B",
"description": "Efficient Indian-language reasoning model for chat, coding, and multilingual work",
"family": "sarvam",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-18",
"last_updated": "2026-02-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0.02,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/sarvam/sarvam-30b\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"sarvam/sarvam-30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5": {
"id": "z-ai/glm-5",
"name": "GLM-5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-11",
"last_updated": "2026-02-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.95,
"output": 3.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/z-ai/glm-5\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"z-ai/glm-5.1": {
"id": "z-ai/glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 1.05,
"output": 3.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fastrouter/z-ai/glm-5.1\", apiKey: processEnvironment[\"FASTROUTER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://go.fastrouter.ai/api/v1\")!,\n apiKey: processEnvironment[\"FASTROUTER_API_KEY\"]\n)\nlet session = provider.model(\"z-ai/glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"cloudflare-ai-gateway": {
"id": "cloudflare-ai-gateway",
"name": "Cloudflare AI Gateway",
"baseURL": "",
"npm": "ai-gateway-provider",
"swiftDriver": "openaiChat",
"env": [
"CLOUDFLARE_API_TOKEN",
"CLOUDFLARE_ACCOUNT_ID",
"CLOUDFLARE_GATEWAY_ID"
],
"doc": "https://developers.cloudflare.com/ai-gateway/",
"modelCount": 44,
"models": {
"alibaba/qwen3.7-max": {
"id": "alibaba/qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 3.75,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/alibaba/qwen3.7-max\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"alibaba/qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.5-397b-a17b": {
"id": "alibaba/qwen3.5-397b-a17b",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/alibaba/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"alibaba/qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3-max": {
"id": "alibaba/qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 1.2,
"output": 6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/alibaba/qwen3-max\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"alibaba/qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.8-max": {
"id": "alibaba/qwen3.8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
},
{
"type": "budget_tokens",
"min": 0,
"max": 262144
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/alibaba/qwen3.8-max\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"alibaba/qwen3.8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"alibaba/qwen3.7-plus": {
"id": "alibaba/qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "budget_tokens"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 0.32,
"output": 1.28,
"cache_read": 0.064
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/alibaba/qwen3.7-plus\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"alibaba/qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.8": {
"id": "anthropic/claude-opus-4.8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/anthropic/claude-opus-4.8\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.7": {
"id": "anthropic/claude-opus-4.7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/anthropic/claude-opus-4.7\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-5": {
"id": "anthropic/claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/anthropic/claude-opus-5\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.6": {
"id": "anthropic/claude-sonnet-4.6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic"
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/anthropic/claude-sonnet-4.6\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-4.5": {
"id": "anthropic/claude-haiku-4.5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"provider": {
"npm": "@ai-sdk/anthropic"
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/anthropic/claude-haiku-4.5\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.6": {
"id": "anthropic/claude-opus-4.6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/anthropic/claude-opus-4.6\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5": {
"id": "anthropic/claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic"
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/anthropic/claude-fable-5\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4.5": {
"id": "anthropic/claude-sonnet-4.5",
"name": "Claude Sonnet 4.5 (latest)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"provider": {
"npm": "@ai-sdk/anthropic"
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/anthropic/claude-sonnet-4.5\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4.5": {
"id": "anthropic/claude-opus-4.5",
"name": "Claude Opus 4.5 (latest)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"provider": {
"npm": "@ai-sdk/anthropic"
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/anthropic/claude-opus-4.5\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-5": {
"id": "anthropic/claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic"
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/anthropic/claude-sonnet-5\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5.1": {
"id": "anthropic/claude-fable-5.1",
"name": "Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/anthropic"
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/anthropic/claude-fable-5.1\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-pro": {
"id": "deepseek/deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 384000
},
"cost": {
"input": 1.74,
"output": 3.48,
"cache_read": 0.145
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/deepseek/deepseek-v4-pro\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-nano": {
"id": "openai/gpt-5-nano",
"name": "GPT-5 Nano",
"description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5-nano\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-nano": {
"id": "openai/gpt-4.1-nano",
"name": "GPT-4.1 nano",
"description": "Tiny GPT-4.1 option for classification, routing, and very high-volume tasks",
"family": "gpt-nano",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-4.1-nano\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-sol": {
"id": "openai/gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.25,
"cache_write": 3.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5.6-sol\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1-mini": {
"id": "openai/gpt-4.1-mini",
"name": "GPT-4.1 mini",
"description": "Affordable GPT-4.1 lane for fast coding help and structured extraction",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-4.1-mini\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-4.1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4": {
"id": "openai/gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5.4\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1": {
"id": "openai/gpt-5.1",
"name": "GPT-5.1",
"description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5.1\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o": {
"id": "openai/gpt-4o",
"name": "GPT-4o",
"description": "Omni-era GPT for multimodal chat, practical coding, and general assistants",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-05-13",
"last_updated": "2024-08-06",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 1.25,
"output": 5,
"cache_read": 0.625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-4o\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-4o\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-luna": {
"id": "openai/gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5.6-luna\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4o-mini": {
"id": "openai/gpt-4o-mini",
"name": "GPT-4o mini",
"description": "Small omni GPT for cheap multimodal assistance and production-scale traffic",
"family": "gpt-mini",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2023-09",
"release_date": "2024-07-18",
"last_updated": "2024-07-18",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 0.075,
"output": 0.3,
"cache_read": 0.0375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-4o-mini\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-4o-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-4.1": {
"id": "openai/gpt-4.1",
"name": "GPT-4.1",
"description": "Long-lived GPT workhorse for coding, instruction following, and production apps",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2025-04-14",
"last_updated": "2025-04-14",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1047576,
"output": 32768
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-4.1\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-4.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-nano": {
"id": "openai/gpt-5.4-nano",
"name": "GPT-5.4 nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5.4-nano\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5-pro": {
"id": "openai/gpt-5.5-pro",
"name": "GPT-5.5 Pro",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 30,
"output": 180
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5.5-pro\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-mini": {
"id": "openai/gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5.4-mini\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-mini": {
"id": "openai/gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5-mini\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-pro": {
"id": "openai/gpt-5.4-pro",
"name": "GPT-5.4 Pro",
"description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 30,
"output": 180
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5.4-pro\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-terra": {
"id": "openai/gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5.6-terra\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5": {
"id": "openai/gpt-5",
"name": "GPT-5",
"description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"input": 272000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o4-mini": {
"id": "openai/o4-mini",
"name": "o4-mini",
"description": "Fast o-series model for compact reasoning, coding, and tool use",
"family": "o-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.275
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/o4-mini\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/o4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3-mini": {
"id": "openai/o3-mini",
"name": "o3-mini",
"description": "Smaller o-series reasoner for economical coding, math, and planning tasks",
"family": "o-mini",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2024-12-20",
"last_updated": "2025-01-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 1.1,
"output": 4.4,
"cache_read": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/o3-mini\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/o3-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/o3": {
"id": "openai/o3",
"name": "o3",
"description": "Deliberate o-series reasoner for hard math, coding, and multi-step analysis",
"family": "o",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05",
"release_date": "2025-04-16",
"last_updated": "2025-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 100000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/o3\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/o3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 922000,
"output": 128000
},
"provider": {
"npm": "@ai-sdk/openai"
},
"cost": {
"input": 5,
"output": 30
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/openai/gpt-5.5\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/kimi-k3": {
"id": "moonshotai/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/moonshotai/kimi-k3\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"moonshotai/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.3": {
"id": "xai/grok-4.3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/xai/grok-4.3\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"xai/grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.20-0309-reasoning": {
"id": "xai/grok-4.20-0309-reasoning",
"name": "Grok 4.20 (Reasoning)",
"description": "Reasoning Grok for document-heavy analysis and long-horizon tool use",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/xai/grok-4.20-0309-reasoning\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"xai/grok-4.20-0309-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.5": {
"id": "xai/grok-4.5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/xai/grok-4.5\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"xai/grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.6": {
"id": "xai/grok-4.6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/xai/grok-4.6\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"xai/grok-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.20-0309-non-reasoning": {
"id": "xai/grok-4.20-0309-non-reasoning",
"name": "Grok 4.20 (Non-Reasoning)",
"description": "Grok model for agentic tool use, reasoning, coding, and live assistance",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-09",
"last_updated": "2026-03-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"cloudflare-ai-gateway/xai/grok-4.20-0309-non-reasoning\", apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"CLOUDFLARE_API_TOKEN\"]\n)\nlet session = provider.model(\"xai/grok-4.20-0309-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"github-copilot": {
"id": "github-copilot",
"name": "GitHub Copilot",
"baseURL": "https://api.githubcopilot.com",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"GITHUB_TOKEN"
],
"doc": "https://docs.github.com/en/copilot",
"modelCount": 28,
"models": {
"claude-opus-4.8": {
"id": "claude-opus-4.8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 168000,
"output": 64000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"provider": {
"body": {
"speed": "fast"
},
"headers": {
"anthropic-beta": "fast-mode-2026-02-01"
}
}
}
}
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/claude-opus-4.8\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"claude-opus-4.8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-sol": {
"id": "gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 4,
"output": 20,
"cache_read": 0.4,
"cache_write": 5,
"tiers": [
{
"input": 8,
"output": 30,
"cache_read": 0.8,
"cache_write": 10,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 8,
"output": 30,
"cache_read": 0.8,
"cache_write": 10
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gpt-5.6-sol\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-4.7": {
"id": "claude-opus-4.7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 168000,
"output": 32000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 30,
"output": 150,
"cache_read": 3,
"cache_write": 37.5
},
"provider": {
"body": {
"speed": "fast"
},
"headers": {
"anthropic-beta": "fast-mode-2026-02-01"
}
}
}
}
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/claude-opus-4.7\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"claude-opus-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-opus-5": {
"id": "claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 936000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/claude-opus-5\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-6-astra": {
"id": "gpt-6-astra",
"name": "GPT-6 Astra",
"description": "GPT-6 Astra is OpenAI's most capable model for complex reasoning, coding, computer use, research, and document creation.",
"family": "gpt-astra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-04-30",
"release_date": "2026-09-04",
"last_updated": "2026-09-04",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5,
"tiers": [
{
"input": 20,
"output": 75,
"cache_read": 2,
"cache_write": 25,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 20,
"output": 75,
"cache_read": 2,
"cache_write": 25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gpt-6-astra\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gpt-6-astra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code": {
"id": "kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 224000,
"output": 32000
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/kimi-k2.7-code\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-4.6": {
"id": "claude-sonnet-4.6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024,
"max": 32000
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 168000,
"output": 32000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/claude-sonnet-4.6\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"claude-sonnet-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.5": {
"id": "grok-4.5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"input": 372000,
"output": 128000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5,
"tiers": [
{
"input": 4,
"output": 12,
"cache_read": 1,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 12,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/grok-4.5\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.6-flash": {
"id": "gemini-3.6-flash",
"name": "Gemini 3.6 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 256,
"max": 32000
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 936000,
"output": 64000
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gemini-3.6-flash\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gemini-3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gpt-5.4\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mai-code-1-flash-picker": {
"id": "mai-code-1-flash-picker",
"name": "MAI-Code-1-Flash",
"description": "Microsoft coding model built for fast, efficient assistance in everyday developer workflows",
"family": "mai",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2026-06-02",
"last_updated": "2026-06-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 128000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/mai-code-1-flash-picker\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"mai-code-1-flash-picker\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-haiku-4.5": {
"id": "claude-haiku-4.5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024,
"max": 32000
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 136000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/claude-haiku-4.5\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"claude-haiku-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.5-flash": {
"id": "gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 256,
"max": 24000
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"input": 128000,
"output": 64000
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15,
"input_audio": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gemini-3.5-flash\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mai-code-1.1-flash": {
"id": "mai-code-1.1-flash",
"name": "MAI-Code-1.1-Flash",
"description": "Microsoft coding model with native vision support, optimized for fast and efficient software development",
"family": "mai",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"release_date": "2026-08-11",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"input": 128000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/mai-code-1.1-flash\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"mai-code-1.1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-luna": {
"id": "gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25,
"tiers": [
{
"input": 0.4,
"output": 1.8,
"cache_read": 0.04,
"cache_write": 0.5,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.04,
"cache_write": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gpt-5.6-luna\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/kimi-k3\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.3-codex": {
"id": "gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gpt-5.3-codex\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5": {
"id": "claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/claude-fable-5\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-nano": {
"id": "gpt-5.4-nano",
"name": "GPT-5.4 nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gpt-5.4-nano\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.4-mini": {
"id": "gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gpt-5.4-mini\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"grok-4.6": {
"id": "grok-4.6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"input": 372000,
"output": 128000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5,
"tiers": [
{
"input": 4,
"output": 12,
"cache_read": 1,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 12,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/grok-4.6\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"grok-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.8-flash": {
"id": "gemini-3.8-flash",
"name": "Gemini 3.8 Flash",
"description": "Google's most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 936000,
"output": 64000
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gemini-3.8-flash\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gemini-3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5-mini": {
"id": "gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 264000,
"input": 128000,
"output": 64000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gpt-5-mini\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini-3.7-flash": {
"id": "gemini-3.7-flash",
"name": "Gemini 3.7 Flash",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 936000,
"output": 64000
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gemini-3.7-flash\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.6-terra": {
"id": "gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gpt-5.6-terra\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-sonnet-5": {
"id": "claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/claude-sonnet-5\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"claude-fable-5.1": {
"id": "claude-fable-5.1",
"name": "Claude Fable 5.1",
"description": "Claude model for demanding reasoning and long-horizon agentic work",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-06",
"release_date": "2026-09-01",
"last_updated": "2026-09-01",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/claude-fable-5.1\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"claude-fable-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-5.5": {
"id": "gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"github-copilot/gpt-5.5\", apiKey: processEnvironment[\"GITHUB_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.githubcopilot.com\")!,\n apiKey: processEnvironment[\"GITHUB_TOKEN\"]\n)\nlet session = provider.model(\"gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"zhipuai": {
"id": "zhipuai",
"name": "Zhipu AI",
"baseURL": "https://open.bigmodel.cn/api/paas/v4",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ZHIPU_API_KEY"
],
"doc": "https://docs.z.ai/guides/overview/pricing",
"modelCount": 15,
"models": {
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-5.2\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.075,
"output": 0.25,
"cache_read": 0.015,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-5.3-flash\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5": {
"id": "glm-5",
"name": "GLM-5",
"description": "General GLM flagship for coding, analysis, and tool-heavy engineering workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-02-11",
"last_updated": "2026-02-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.2,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-5\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-27",
"last_updated": "2026-03-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-5.1\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-5.3\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5v-turbo": {
"id": "glm-5v-turbo",
"name": "GLM-5V-Turbo",
"description": "Fast GLM vision model for screenshots, documents, and multimodal agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-04-01",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 5,
"output": 22,
"cache_read": 1.2,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-5v-turbo\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-5v-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.7-flash": {
"id": "glm-4.7-flash",
"name": "GLM-4.7-Flash",
"description": "Budget GLM lane for fast coding help, routing, and everyday automation",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-4.7-flash\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.7-flashx": {
"id": "glm-4.7-flashx",
"name": "GLM-4.7-FlashX",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0.07,
"output": 0.4,
"cache_read": 0.01,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-4.7-flashx\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.7-flashx\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.5v": {
"id": "glm-4.5v",
"name": "GLM-4.5V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-08-11",
"last_updated": "2025-08-11",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 64000,
"output": 16384
},
"cost": {
"input": 0.6,
"output": 1.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-4.5v\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.5": {
"id": "glm-4.5",
"name": "GLM-4.5",
"description": "Hybrid-reasoning GLM release that made the 4.5 line broadly useful",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-4.5\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.5-flash": {
"id": "glm-4.5-flash",
"name": "GLM-4.5-Flash",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-4.5-flash\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.6v": {
"id": "glm-4.6v",
"name": "GLM-4.6V",
"description": "GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 0.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-4.6v\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.6v\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.6": {
"id": "glm-4.6",
"name": "GLM-4.6",
"description": "Late GLM-4 workhorse for coding agents, reasoning, and structured tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-4.6\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.5-air": {
"id": "glm-4.5-air",
"name": "GLM-4.5-Air",
"description": "Lighter GLM-4.5 variant for fast coding assistance and cheaper agents",
"family": "glm-air",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 98304
},
"cost": {
"input": 0.2,
"output": 1.1,
"cache_read": 0.03,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-4.5-air\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.5-air\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4.7": {
"id": "glm-4.7",
"name": "GLM-4.7",
"description": "Mature GLM model for dependable coding, reasoning, and structured agent tasks",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0.6,
"output": 2.2,
"cache_read": 0.11,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"zhipuai/glm-4.7\", apiKey: processEnvironment[\"ZHIPU_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://open.bigmodel.cn/api/paas/v4\")!,\n apiKey: processEnvironment[\"ZHIPU_API_KEY\"]\n)\nlet session = provider.model(\"glm-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"jalapeno": {
"id": "jalapeno",
"name": "Jalapeno Cloud",
"baseURL": "https://api.jalapeno-cloud.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"JALAPENO_API_KEY"
],
"doc": "https://www.jalapeno-cloud.ai/docs/",
"modelCount": 17,
"models": {
"Qwen3.5-27B": {
"id": "Qwen3.5-27B",
"name": "Qwen3.5 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Qwen3.5-27B\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3.5-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"DeepSeek-V4-Flash": {
"id": "DeepSeek-V4-Flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.14,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/DeepSeek-V4-Flash\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"DeepSeek-V4-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"GLM-5.1": {
"id": "GLM-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 1.38,
"output": 4.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/GLM-5.1\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"GLM-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3.5-122B-A10B": {
"id": "Qwen3.5-122B-A10B",
"name": "Qwen3.5 122B-A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Qwen3.5-122B-A10B\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3.5-122B-A10B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3-VL-235B-A22B-Instruct": {
"id": "Qwen3-VL-235B-A22B-Instruct",
"name": "Qwen3 VL 235B A22B Instruct",
"description": "Qwen vision-language instruct model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 129024,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Qwen3-VL-235B-A22B-Instruct\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3-VL-235B-A22B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Kimi-K2.5": {
"id": "Kimi-K2.5",
"name": "Kimi K2.5",
"description": "Earlier Kimi frontier model for long-context agents, coding, and multimodal work",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-01",
"last_updated": "2026-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 180224
},
"cost": {
"input": 0.6,
"output": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Kimi-K2.5\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Kimi-K2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Kimi-K2.7-Code": {
"id": "Kimi-K2.7-Code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 271360,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Kimi-K2.7-Code\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Kimi-K2.7-Code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3-Next-80B-A3B-Instruct": {
"id": "Qwen3-Next-80B-A3B-Instruct",
"name": "Qwen3-Next 80B-A3B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 129024,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Qwen3-Next-80B-A3B-Instruct\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3-Next-80B-A3B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3.5-397B-A17B": {
"id": "Qwen3.5-397B-A17B",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.6,
"output": 3.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Qwen3.5-397B-A17B\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3.5-397B-A17B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3.5-35B-A3B": {
"id": "Qwen3.5-35B-A3B",
"name": "Qwen3.5 35B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Qwen3.5-35B-A3B\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3.5-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"GLM-5.2": {
"id": "GLM-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/GLM-5.2\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M3": {
"id": "MiniMax-M3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 512000
},
"cost": {
"input": 0.3,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/MiniMax-M3\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Hy3": {
"id": "Hy3",
"name": "Hy3",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"input": 192000,
"output": 128000
},
"cost": {
"input": 0.14,
"output": 0.58
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Hy3\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3-VL-235B-A22B-Thinking": {
"id": "Qwen3-VL-235B-A22B-Thinking",
"name": "Qwen3 VL 235B A22B Thinking",
"description": "Qwen vision-language thinking model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.98,
"output": 3.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Qwen3-VL-235B-A22B-Thinking\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3-VL-235B-A22B-Thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3-Next-80B-A3B-Thinking": {
"id": "Qwen3-Next-80B-A3B-Thinking",
"name": "Qwen3-Next 80B-A3B (Thinking)",
"description": "Efficient Qwen thinking model for local reasoning, math, and coding agents",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Qwen3-Next-80B-A3B-Thinking\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3-Next-80B-A3B-Thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Kimi-K3": {
"id": "Kimi-K3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/Kimi-K3\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"Kimi-K3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"DeepSeek-V4-Pro": {
"id": "DeepSeek-V4-Pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 1.6,
"output": 3.38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"jalapeno/DeepSeek-V4-Pro\", apiKey: processEnvironment[\"JALAPENO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.jalapeno-cloud.ai/v1\")!,\n apiKey: processEnvironment[\"JALAPENO_API_KEY\"]\n)\nlet session = provider.model(\"DeepSeek-V4-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"perplexity-agent": {
"id": "perplexity-agent",
"name": "Perplexity Agent",
"baseURL": "https://api.perplexity.ai/v1",
"npm": "@ai-sdk/openai",
"swiftDriver": "openaiChat",
"env": [
"PERPLEXITY_API_KEY"
],
"doc": "https://docs.perplexity.ai/docs/agent-api/models",
"modelCount": 22,
"models": {
"nvidia/nemotron-3-super-120b-a12b": {
"id": "nvidia/nemotron-3-super-120b-a12b",
"name": "Nemotron 3 Super 120B",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2026-02",
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 32000
},
"cost": {
"input": 0.25,
"output": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/nvidia/nemotron-3-super-120b-a12b\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/nemotron-3-super-120b-a12b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4-6": {
"id": "anthropic/claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-02-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/anthropic/claude-sonnet-4-6\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-5": {
"id": "anthropic/claude-opus-4-5",
"name": "Claude Opus 4.5",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/anthropic/claude-opus-4-5\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-6": {
"id": "anthropic/claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/anthropic/claude-opus-4-6\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-7": {
"id": "anthropic/claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/anthropic/claude-opus-4-7\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-4-5": {
"id": "anthropic/claude-haiku-4-5",
"name": "Claude Haiku 4.5",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/anthropic/claude-haiku-4-5\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4-5": {
"id": "anthropic/claude-sonnet-4-5",
"name": "Claude Sonnet 4.5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/anthropic/claude-sonnet-4-5\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3.1-pro-preview": {
"id": "google/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/google/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-3-flash-preview": {
"id": "google/gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"tiers": [
{
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/google/gemini-3-flash-preview\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-pro": {
"id": "google/gemini-2.5-pro",
"name": "Gemini 2.5 Pro",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-03-20",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/google/gemini-2.5-pro\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemini-2.5-flash": {
"id": "google/gemini-2.5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-03-20",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/google/gemini-2.5-flash\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"google/gemini-2.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot-ai/kimi-k2.7-code": {
"id": "moonshot-ai/kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/moonshot-ai/kimi-k2.7-code\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"moonshot-ai/kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot-ai/kimi-k3": {
"id": "moonshot-ai/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/moonshot-ai/kimi-k3\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"moonshot-ai/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek/deepseek-v4-flash-0731": {
"id": "deepseek/deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.13,
"output": 0.26,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/deepseek/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"deepseek/deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4": {
"id": "openai/gpt-5.4",
"name": "GPT-5.4",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/openai/gpt-5.4\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.1": {
"id": "openai/gpt-5.1",
"name": "GPT-5.1",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/openai/gpt-5.1\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5-mini": {
"id": "openai/gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Compact GPT model for low-latency assistance and high-volume workloads",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/openai/gpt-5-mini\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.2": {
"id": "openai/gpt-5.2",
"name": "GPT-5.2",
"description": "GPT model for general reasoning, writing, coding, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/openai/gpt-5.2\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT-5.5",
"description": "Frontier GPT model for professional reasoning, coding, and multimodal work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/openai/gpt-5.5\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4-1-fast-non-reasoning": {
"id": "xai/grok-4-1-fast-non-reasoning",
"name": "Grok 4.1 Fast (Non-Reasoning)",
"description": "Fast Grok model for responsive chat, reasoning, and tool-assisted work",
"family": "grok",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-07",
"release_date": "2025-11-19",
"last_updated": "2025-11-19",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 2000000,
"output": 30000
},
"cost": {
"input": 0.2,
"output": 0.5,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/xai/grok-4-1-fast-non-reasoning\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4-1-fast-non-reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.6": {
"id": "xai/grok-4.6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5,
"tiers": [
{
"input": 4,
"output": 12,
"cache_read": 1,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 12,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/xai/grok-4.6\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/sonar": {
"id": "perplexity/sonar",
"name": "Sonar",
"description": "Sonar search model for current answers, retrieval, and citation-backed chat",
"family": "sonar",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0.25,
"output": 2.5,
"cache_read": 0.0625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"perplexity-agent/perplexity/sonar\", apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.perplexity.ai/v1\")!,\n apiKey: processEnvironment[\"PERPLEXITY_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/sonar\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"fireworks-ai": {
"id": "fireworks-ai",
"name": "Fireworks AI",
"baseURL": "https://api.fireworks.ai/inference/v1/",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"FIREWORKS_API_KEY"
],
"doc": "https://fireworks.ai/docs/",
"modelCount": 23,
"models": {
"accounts/fireworks/routers/kimi-k3-fast": {
"id": "accounts/fireworks/routers/kimi-k3-fast",
"name": "Kimi K3 Fast",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-27",
"last_updated": "2026-07-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 4.5,
"output": 22.5,
"cache_read": 0.45
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/routers/kimi-k3-fast\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/routers/kimi-k3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/routers/glm-5p3-fast": {
"id": "accounts/fireworks/routers/glm-5p3-fast",
"name": "GLM 5.3 Fast",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-28",
"last_updated": "2026-09-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048572,
"output": 262144
},
"cost": {
"input": 2.1,
"output": 6.6,
"cache_read": 0.39
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/routers/glm-5p3-fast\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/routers/glm-5p3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/routers/glm-5p2-fast": {
"id": "accounts/fireworks/routers/glm-5p2-fast",
"name": "GLM 5.2 Fast",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-06-26",
"last_updated": "2026-06-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048575,
"output": 131072
},
"cost": {
"input": 2.1,
"output": 6.6,
"cache_read": 0.21
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/routers/glm-5p2-fast\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/routers/glm-5p2-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/qwen3p7-plus": {
"id": "accounts/fireworks/models/qwen3p7-plus",
"name": "Qwen 3.7 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
},
{
"type": "budget_tokens",
"min": 1
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.08
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/qwen3p7-plus\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/qwen3p7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/deepseek-v4-flash-vision-exp": {
"id": "accounts/fireworks/models/deepseek-v4-flash-vision-exp",
"name": "DeepSeek V4 Flash Vision Exp",
"description": "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-21",
"last_updated": "2026-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.22,
"output": 0.66,
"cache_read": 0.007
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/deepseek-v4-flash-vision-exp\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/deepseek-v4-flash-vision-exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/deepseek-v4-pro-0813": {
"id": "accounts/fireworks/models/deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 1.32,
"output": 3.96,
"cache_read": 0.044
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/deepseek-v4-flash-0731": {
"id": "accounts/fireworks/models/deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.22,
"output": 0.66,
"cache_read": 0.007
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/minimax-m3": {
"id": "accounts/fireworks/models/minimax-m3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal coding model for long-context reasoning and agent tasks",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 512000,
"output": 512000
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/minimax-m3\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/deepseek-v4p1-flash": {
"id": "accounts/fireworks/models/deepseek-v4p1-flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 384000
},
"cost": {
"input": 0.22,
"output": 0.66,
"cache_read": 0.007
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/deepseek-v4p1-flash\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/deepseek-v4p1-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/kimi-k2p6": {
"id": "accounts/fireworks/models/kimi-k2p6",
"name": "Kimi K2.6",
"description": "Kimi reasoning model for long-horizon research, planning, and tool use",
"family": "kimi-thinking",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/kimi-k2p6\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/kimi-k2p6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/nemotron-3-ultra-nvfp4": {
"id": "accounts/fireworks/models/nemotron-3-ultra-nvfp4",
"name": "Nemotron 3 Ultra 550B A55B",
"description": "Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-04",
"last_updated": "2026-06-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 128000
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.119
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/nemotron-3-ultra-nvfp4\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/nemotron-3-ultra-nvfp4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/kimi-k3": {
"id": "accounts/fireworks/models/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-27",
"last_updated": "2026-07-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/kimi-k3\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/glm-5p3": {
"id": "accounts/fireworks/models/glm-5p3",
"name": "GLM 5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-09-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048573,
"output": 262144
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/glm-5p3\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/glm-5p3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/kimi-k2p7-code": {
"id": "accounts/fireworks/models/kimi-k2p7-code",
"name": "Kimi K2.7 Code",
"description": "Kimi coding model for software agents, refactors, and repository reasoning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-06-12",
"last_updated": "2026-06-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 262000
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/kimi-k2p7-code\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/kimi-k2p7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/glm-5p3-flash": {
"id": "accounts/fireworks/models/glm-5p3-flash",
"name": "GLM 5.3 Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-09-07",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048573,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/glm-5p3-flash\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/glm-5p3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/qwen3p8-2p4t-a95b": {
"id": "accounts/fireworks/models/qwen3p8-2p4t-a95b",
"name": "Qwen3.8 2.4T A95B",
"description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/qwen3p8-2p4t-a95b\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/qwen3p8-2p4t-a95b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/muse-glimmer-30b": {
"id": "accounts/fireworks/models/muse-glimmer-30b",
"name": "Muse Glimmer 30B",
"description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-01-04",
"release_date": "2026-08-10",
"last_updated": "2026-08-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.35,
"output": 1.5,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/muse-glimmer-30b\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/muse-glimmer-30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/inkling": {
"id": "accounts/fireworks/models/inkling",
"name": "Inkling",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 1048576
},
"cost": {
"input": 1,
"output": 4.05,
"cache_read": 0.17
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/inkling\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/gpt-oss-120b": {
"id": "accounts/fireworks/models/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2026-06-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/gpt-oss-120b\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/mistral-large-3-fp8": {
"id": "accounts/fireworks/models/mistral-large-3-fp8",
"name": "Mistral Large 3 675B Instruct 2512",
"description": "Mistral's largest general model for enterprise agents, coding, and multilingual reasoning",
"family": "mistral-large",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-11",
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/mistral-large-3-fp8\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/mistral-large-3-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/glm-5p2": {
"id": "accounts/fireworks/models/glm-5p2",
"name": "GLM 5.2",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-06-16",
"last_updated": "2026-06-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048575,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/glm-5p2\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/glm-5p2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/qwen3p8-max": {
"id": "accounts/fireworks/models/qwen3p8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/qwen3p8-max\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/qwen3p8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"accounts/fireworks/models/nemotron-lightning-3p5-30b-a3b": {
"id": "accounts/fireworks/models/nemotron-lightning-3p5-30b-a3b",
"name": "Nemotron 3.5 Lightning 30B A3B",
"description": "Fast NVIDIA Nemotron MoE for reliable agentic tasks across enterprise workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-11",
"last_updated": "2026-08-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.05,
"output": 0.2,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"fireworks-ai/accounts/fireworks/models/nemotron-lightning-3p5-30b-a3b\", apiKey: processEnvironment[\"FIREWORKS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.fireworks.ai/inference/v1/\")!,\n apiKey: processEnvironment[\"FIREWORKS_API_KEY\"]\n)\nlet session = provider.model(\"accounts/fireworks/models/nemotron-lightning-3p5-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"opper": {
"id": "opper",
"name": "Opper",
"baseURL": "https://api.opper.ai/v3/compat",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"OPPER_API_KEY"
],
"doc": "https://opper.ai/models",
"modelCount": 40,
"models": {
"minimax/m3": {
"id": "minimax/m3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 512000
},
"cost": {
"input": 0.6,
"output": 2.4,
"cache_read": 0.12,
"tiers": [
{
"input": 1.2,
"output": 4.8,
"cache_read": 0.24,
"tier": {
"type": "context",
"size": 524288
}
}
],
"context_over_200k": {
"input": 1.2,
"output": 4.8,
"cache_read": 0.24
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/minimax/m3\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"minimax/m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4-6": {
"id": "anthropic/claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/anthropic/claude-sonnet-4-6\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-5": {
"id": "anthropic/claude-opus-5",
"name": "Claude Opus 5",
"description": "Strongest Claude Opus model for coding, agents, and professional work",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-05",
"release_date": "2026-07-24",
"last_updated": "2026-07-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/anthropic/claude-opus-5\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-5": {
"id": "anthropic/claude-opus-4-5",
"name": "Claude Opus 4.5 (latest)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/anthropic/claude-opus-4-5\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-6": {
"id": "anthropic/claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/anthropic/claude-opus-4-6\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-7": {
"id": "anthropic/claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/anthropic/claude-opus-4-7\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-fable-5": {
"id": "anthropic/claude-fable-5",
"name": "Claude Fable 5",
"description": "Claude model for creative writing, analysis, and controlled agent workflows",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 1,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/anthropic/claude-fable-5\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-fable-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-haiku-4-5": {
"id": "anthropic/claude-haiku-4-5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/anthropic/claude-haiku-4-5\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-4-5": {
"id": "anthropic/claude-sonnet-4-5",
"name": "Claude Sonnet 4.5 (latest)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/anthropic/claude-sonnet-4-5\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-opus-4-8": {
"id": "anthropic/claude-opus-4-8",
"name": "Claude Opus 4.8",
"description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-01",
"release_date": "2026-05-28",
"last_updated": "2026-05-28",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/anthropic/claude-opus-4-8\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-opus-4-8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"anthropic/claude-sonnet-5": {
"id": "anthropic/claude-sonnet-5",
"name": "Claude Sonnet 5",
"description": "Everyday Claude agent model for coding, planning, browsing, and general work",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-06-30",
"last_updated": "2026-06-30",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/anthropic/claude-sonnet-5\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"anthropic/claude-sonnet-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta/muse-spark-1.2": {
"id": "meta/muse-spark-1.2",
"name": "Muse Spark 1.2",
"description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/meta/muse-spark-1.2\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"meta/muse-spark-1.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshot/kimi-k3": {
"id": "moonshot/kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/moonshot/kimi-k3\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"moonshot/kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-3.1-pro-preview": {
"id": "gemini/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"description": "Reasoning-first Gemini preview for agentic coding and complex problem solving",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/gemini/gemini-3.1-pro-preview\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-3.1-pro-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-3.5-flash": {
"id": "gemini/gemini-3.5-flash",
"name": "Gemini 3.5 Flash",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-05-19",
"last_updated": "2026-05-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/gemini/gemini-3.5-flash\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-3.5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-3.5-flash-lite": {
"id": "gemini/gemini-3.5-flash-lite",
"name": "Gemini 3.5 Flash Lite",
"description": "Fast Gemini model balancing multimodal reasoning, tool use, and cost",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/gemini/gemini-3.5-flash-lite\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-3.5-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemini/gemini-3-flash-preview": {
"id": "gemini/gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/gemini/gemini-3-flash-preview\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"gemini/gemini-3-flash-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-sol": {
"id": "openai/gpt-5.6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"cache_write": 6.25,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"cache_write": 12.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1,
"cache_write": 12.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/openai/gpt-5.6-sol\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4": {
"id": "openai/gpt-5.4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/openai/gpt-5.4\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-luna": {
"id": "openai/gpt-5.6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.2,
"cache_read": 0.02,
"cache_write": 0.25,
"tiers": [
{
"input": 0.4,
"output": 1.8,
"cache_read": 0.04,
"cache_write": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 0.4,
"output": 1.8,
"cache_read": 0.04,
"cache_write": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/openai/gpt-5.6-luna\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-codex": {
"id": "openai/gpt-5.3-codex",
"name": "GPT-5.3 Codex",
"description": "Coding-optimized GPT model for repository edits, reviews, and agentic software work",
"family": "gpt-codex",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/openai/gpt-5.3-codex\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-codex\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-nano": {
"id": "openai/gpt-5.4-nano",
"name": "GPT-5.4 nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/openai/gpt-5.4-nano\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5-pro": {
"id": "openai/gpt-5.5-pro",
"name": "GPT-5.5 Pro",
"description": "Highest-accuracy GPT-5.5 tier for slower, precision-heavy reasoning and coding",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180,
"tiers": [
{
"input": 60,
"output": 270,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 60,
"output": 270
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/openai/gpt-5.5-pro\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-mini": {
"id": "openai/gpt-5.4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/openai/gpt-5.4-mini\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.4-pro": {
"id": "openai/gpt-5.4-pro",
"name": "GPT-5.4 Pro",
"description": "More exact GPT-5.4 tier for demanding professional reasoning and agent tasks",
"family": "gpt-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 30,
"output": 180,
"tiers": [
{
"input": 60,
"output": 270,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 60,
"output": 270
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/openai/gpt-5.4-pro\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.6-terra": {
"id": "openai/gpt-5.6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"cache_write": 2.5,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4,
"cache_write": 5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/openai/gpt-5.6-terra\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.3-chat-latest": {
"id": "openai/gpt-5.3-chat-latest",
"name": "GPT-5.3 Chat (latest)",
"description": "Chat-tuned GPT model for conversational assistance, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 16384
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/openai/gpt-5.3-chat-latest\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.3-chat-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-5.5": {
"id": "openai/gpt-5.5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/openai/gpt-5.5\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-5.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.3": {
"id": "xai/grok-4.3",
"name": "Grok 4.3",
"description": "xAI's default Grok for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 30000
},
"cost": {
"input": 1.25,
"output": 2.5,
"cache_read": 0.2,
"tiers": [
{
"input": 2.5,
"output": 5,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 5,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/xai/grok-4.3\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.5": {
"id": "xai/grok-4.5",
"name": "Grok 4.5",
"description": "xAI's Grok model for chat, coding, agentic tools, and lower hallucination risk",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.3,
"tiers": [
{
"input": 4,
"output": 12,
"cache_read": 0.6,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 12,
"cache_read": 0.6
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/xai/grok-4.5\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-build-0.1": {
"id": "xai/grok-build-0.1",
"name": "Grok Build 0.1",
"description": "Fast Grok coding model tuned for agentic engineering and iterative edits",
"family": "grok-build",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 1,
"output": 2,
"cache_read": 0.2,
"tiers": [
{
"input": 2,
"output": 4,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2,
"output": 4,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/xai/grok-build-0.1\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-build-0.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"xai/grok-4.6": {
"id": "xai/grok-4.6",
"name": "Grok 4.6",
"description": "xAI's frontier model for long-running agents, coding, knowledge work, and visual projects",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2026-02-01",
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 500000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5,
"tiers": [
{
"input": 4,
"output": 12,
"cache_read": 1,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 12,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/xai/grok-4.6\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"xai/grok-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/devstral-2512": {
"id": "mistral/devstral-2512",
"name": "Devstral 2",
"description": "Mistral's coding-agent model for repository work, terminal tasks, and software fixes",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2025-12-09",
"last_updated": "2025-12-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.4,
"output": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/mistral/devstral-2512\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"mistral/devstral-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-large-2512": {
"id": "mistral/mistral-large-2512",
"name": "Mistral Large 3",
"description": "Mistral's largest general model for enterprise agents, coding, and multilingual reasoning",
"family": "mistral-large",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2024-11",
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.5,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/mistral/mistral-large-2512\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-large-2512\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral/mistral-small-2603": {
"id": "mistral/mistral-small-2603",
"name": "Mistral Small 4",
"description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents",
"family": "mistral-small",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 256000
},
"cost": {
"input": 0.15,
"output": 0.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/mistral/mistral-small-2603\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"mistral/mistral-small-2603\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertexai/gemini-3.7-flash-eu": {
"id": "vertexai/gemini-3.7-flash-eu",
"name": "Gemini 3.7 Flash (EU)",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/vertexai/gemini-3.7-flash-eu\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"vertexai/gemini-3.7-flash-eu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"vertexai/gemini-3.7-flash": {
"id": "vertexai/gemini-3.7-flash",
"name": "Gemini 3.7 Flash",
"description": "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03",
"release_date": "2026-08-13",
"last_updated": "2026-08-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.75,
"output": 3.75,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/vertexai/gemini-3.7-flash\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"vertexai/gemini-3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/sonar": {
"id": "perplexity/sonar",
"name": "Sonar",
"description": "Fast web-grounded Sonar for current answers, citations, and lightweight retrieval",
"family": "sonar",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 1,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/perplexity/sonar\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/sonar\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/sonar-reasoning-pro": {
"id": "perplexity/sonar-reasoning-pro",
"name": "Sonar Reasoning Pro",
"description": "Web-grounded Sonar for multi-step research questions that need cited reasoning",
"family": "sonar-reasoning",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": false,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 2,
"output": 8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/perplexity/sonar-reasoning-pro\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/sonar-reasoning-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"perplexity/sonar-pro": {
"id": "perplexity/sonar-pro",
"name": "Sonar Pro",
"description": "Deeper Sonar search model with broader retrieval and stronger synthesis",
"family": "sonar-pro",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"knowledge": "2025-09-01",
"release_date": "2024-01-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 8192
},
"cost": {
"input": 3,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"opper/perplexity/sonar-pro\", apiKey: processEnvironment[\"OPPER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.opper.ai/v3/compat\")!,\n apiKey: processEnvironment[\"OPPER_API_KEY\"]\n)\nlet session = provider.model(\"perplexity/sonar-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"stackit": {
"id": "stackit",
"name": "STACKIT",
"baseURL": "https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"STACKIT_API_KEY"
],
"doc": "https://docs.stackit.cloud/products/data-and-ai/ai-model-serving/basics/available-shared-models",
"modelCount": 8,
"models": {
"google/gemma-3-27b-it": {
"id": "google/gemma-3-27b-it",
"name": "Gemma 3 27B",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-05-17",
"last_updated": "2025-05-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 37000,
"output": 4096
},
"cost": {
"input": 0.53,
"output": 0.76
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stackit/google/gemma-3-27b-it\", apiKey: processEnvironment[\"STACKIT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1\")!,\n apiKey: processEnvironment[\"STACKIT_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3-27b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-Embedding-8B": {
"id": "Qwen/Qwen3-VL-Embedding-8B",
"name": "Qwen3-VL Embedding 8B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": false,
"release_date": "2026-02-05",
"last_updated": "2026-02-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 4096
},
"cost": {
"input": 0.09,
"output": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stackit/Qwen/Qwen3-VL-Embedding-8B\", apiKey: processEnvironment[\"STACKIT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1\")!,\n apiKey: processEnvironment[\"STACKIT_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-Embedding-8B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-235B-A22B-Instruct-FP8": {
"id": "Qwen/Qwen3-VL-235B-A22B-Instruct-FP8",
"name": "Qwen3-VL 235B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2024-11-01",
"last_updated": "2024-11-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 218000,
"output": 16384
},
"cost": {
"input": 1.76,
"output": 2.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stackit/Qwen/Qwen3-VL-235B-A22B-Instruct-FP8\", apiKey: processEnvironment[\"STACKIT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1\")!,\n apiKey: processEnvironment[\"STACKIT_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-235B-A22B-Instruct-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-27B": {
"id": "Qwen/Qwen3.6-27B",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.53,
"output": 0.76
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stackit/Qwen/Qwen3.6-27B\", apiKey: processEnvironment[\"STACKIT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1\")!,\n apiKey: processEnvironment[\"STACKIT_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"intfloat/e5-mistral-7b-instruct": {
"id": "intfloat/e5-mistral-7b-instruct",
"name": "E5 Mistral 7B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "mistral",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": false,
"release_date": "2023-12-11",
"last_updated": "2023-12-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 4096,
"output": 4096
},
"cost": {
"input": 0.02,
"output": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stackit/intfloat/e5-mistral-7b-instruct\", apiKey: processEnvironment[\"STACKIT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1\")!,\n apiKey: processEnvironment[\"STACKIT_API_KEY\"]\n)\nlet session = provider.model(\"intfloat/e5-mistral-7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.18,
"output": 0.29
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stackit/openai/gpt-oss-20b\", apiKey: processEnvironment[\"STACKIT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1\")!,\n apiKey: processEnvironment[\"STACKIT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131000,
"output": 8192
},
"cost": {
"input": 0.53,
"output": 0.76
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stackit/openai/gpt-oss-120b\", apiKey: processEnvironment[\"STACKIT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1\")!,\n apiKey: processEnvironment[\"STACKIT_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"cortecs/Llama-3.3-70B-Instruct-FP8-Dynamic": {
"id": "cortecs/Llama-3.3-70B-Instruct-FP8-Dynamic",
"name": "Llama 3.3 70B",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2024-12-05",
"last_updated": "2024-12-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.53,
"output": 0.76
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"stackit/cortecs/Llama-3.3-70B-Instruct-FP8-Dynamic\", apiKey: processEnvironment[\"STACKIT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.openai-compat.model-serving.eu01.onstackit.cloud/v1\")!,\n apiKey: processEnvironment[\"STACKIT_API_KEY\"]\n)\nlet session = provider.model(\"cortecs/Llama-3.3-70B-Instruct-FP8-Dynamic\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"crof": {
"id": "crof",
"name": "CrofAI",
"baseURL": "https://crof.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"CROF_API_KEY"
],
"doc": "https://crof.ai/docs",
"modelCount": 24,
"models": {
"greg-2-super": {
"id": "greg-2-super",
"name": "Greg 2 Super",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-06-14",
"last_updated": "2026-06-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 229376,
"output": 229376
},
"cost": {
"input": 1.5,
"output": 5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/greg-2-super\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"greg-2-super\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-vision-exp": {
"id": "deepseek-v4-flash-vision-exp",
"name": "DeepSeek V4 Flash Vision Exp",
"description": "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-21",
"last_updated": "2026-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.08,
"output": 0.2,
"cache_read": 0.007
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/deepseek-v4-flash-vision-exp\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-vision-exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro-0813": {
"id": "deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro (0813)",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.35,
"output": 0.8,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash (New)",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.08,
"output": 0.1,
"cache_read": 0.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-9b": {
"id": "qwen3.5-9b",
"name": "Qwen3.5 9B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-03-13",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.04,
"output": 0.15,
"cache_read": 0.008
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/qwen3.5-9b\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.8-27b": {
"id": "qwen3.8-27b",
"name": "Qwen3.8 27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.2,
"output": 1.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/qwen3.8-27b\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.5,
"output": 1.99,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/kimi-k2.6\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.2": {
"id": "glm-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.3,
"output": 1.05,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/glm-5.2\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.12,
"output": 0.21,
"cache_read": 0.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/deepseek-v4-flash\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.7-code": {
"id": "kimi-k2.7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.55,
"output": 2.25,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/kimi-k2.7-code\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"greg-1-mini": {
"id": "greg-1-mini",
"name": "Greg 1 Mini",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"attachment": true,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 229376,
"output": 229376
},
"cost": {
"input": 0.07,
"output": 0.15,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/greg-1-mini\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"greg-1-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"greg-2-ultra": {
"id": "greg-2-ultra",
"name": "Greg 2 Ultra",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-06-14",
"last_updated": "2026-06-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 229376,
"output": 229376
},
"cost": {
"input": 3,
"output": 10,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/greg-2-ultra\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"greg-2-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.5-397b-a17b": {
"id": "qwen3.5-397b-a17b",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.35,
"output": 1.75,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3.6-27b": {
"id": "qwen3.6-27b",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.2,
"output": 1.5,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/qwen3.6-27b\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"qwen3.6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 262144
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 2,
"output": 8,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/kimi-k3\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3.2": {
"id": "deepseek-v3.2",
"name": "DeepSeek V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-07-22",
"last_updated": "2025-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 0.18,
"output": 0.35,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/deepseek-v3.2\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3-flash": {
"id": "glm-5.3-flash",
"name": "GLM 5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.07,
"output": 0.22,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/glm-5.3-flash\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"greg-rp": {
"id": "greg-rp",
"name": "Greg (Roleplay)",
"description": "General-purpose chat model for instruction following, writing, and analysis",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 229376,
"output": 229376
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/greg-rp\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"greg-rp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-31b-it": {
"id": "gemma-4-31b-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/gemma-4-31b-it\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.1": {
"id": "glm-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 202752
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.45,
"output": 2.15,
"cache_read": 0.08,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/glm-5.1\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3-eco": {
"id": "kimi-k3-eco",
"name": "Kimi K3 Eco",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 1,
"output": 4,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/kimi-k3-eco\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3-eco\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.35,
"output": 0.8,
"cache_read": 0.003
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/deepseek-v4-pro\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.4,
"output": 1.4,
"cache_read": 0.06
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/glm-5.3\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2.5-pro": {
"id": "mimo-v2.5-pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"provider": {
"npm": "@ai-sdk/openai-compatible"
},
"cost": {
"input": 0.4,
"output": 0.8,
"cache_read": 0.003,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crof/mimo-v2.5-pro\", apiKey: processEnvironment[\"CROF_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://crof.ai/v1\")!,\n apiKey: processEnvironment[\"CROF_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2.5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"crusoe": {
"id": "crusoe",
"name": "Crusoe",
"baseURL": "https://api.inference.crusoecloud.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"CRUSOE_API_KEY"
],
"doc": "https://docs.crusoecloud.com/managed-inference/overview",
"modelCount": 11,
"models": {
"deepseek-ai/DeepSeek-V3-0324": {
"id": "deepseek-ai/DeepSeek-V3-0324",
"name": "DeepSeek V3 0324",
"description": "March 2025 checkpoint of DeepSeek-V3 with improved reasoning and coding",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-03-24",
"last_updated": "2025-03-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 0.5,
"output": 1.5,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crusoe/deepseek-ai/DeepSeek-V3-0324\", apiKey: processEnvironment[\"CRUSOE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.crusoecloud.com/v1\")!,\n apiKey: processEnvironment[\"CRUSOE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3-0324\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B": {
"id": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B",
"name": "Nemotron 3 Nano 30B A3B",
"description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-15",
"last_updated": "2025-12-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.05,
"output": 0.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crusoe/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B\", apiKey: processEnvironment[\"CRUSOE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.crusoecloud.com/v1\")!,\n apiKey: processEnvironment[\"CRUSOE_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Nemotron-3-Nano-Omni-Reasoning-30B-A3B": {
"id": "nvidia/Nemotron-3-Nano-Omni-Reasoning-30B-A3B",
"name": "Nemotron 3 Nano Omni 30B A3B Reasoning",
"description": "Open Nemotron omni model combining reasoning with text, vision, and audio",
"family": "nemotron",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"temperature": true,
"release_date": "2026-04-28",
"last_updated": "2026-04-28",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.83,
"cache_read": 0.3,
"input_audio": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crusoe/nvidia/Nemotron-3-Nano-Omni-Reasoning-30B-A3B\", apiKey: processEnvironment[\"CRUSOE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.crusoecloud.com/v1\")!,\n apiKey: processEnvironment[\"CRUSOE_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Nemotron-3-Nano-Omni-Reasoning-30B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/NVIDIA-Nemotron-3-Super-120B-A12B": {
"id": "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B",
"name": "Nemotron 3 Super 120B A12B",
"description": "Nemotron middle tier for collaborative agents and high-volume reasoning workloads",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-11",
"last_updated": "2026-03-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.3,
"output": 2.4,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crusoe/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B\", apiKey: processEnvironment[\"CRUSOE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.crusoecloud.com/v1\")!,\n apiKey: processEnvironment[\"CRUSOE_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/NVIDIA-Nemotron-3-Super-120B-A12B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31b-it": {
"id": "google/gemma-4-31b-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.14,
"output": 0.4,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crusoe/google/gemma-4-31b-it\", apiKey: processEnvironment[\"CRUSOE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.crusoecloud.com/v1\")!,\n apiKey: processEnvironment[\"CRUSOE_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B-Instruct-2507": {
"id": "Qwen/Qwen3-235B-A22B-Instruct-2507",
"name": "Qwen3 235B-A22B Instruct 2507",
"description": "Updated large open Qwen3 MoE instruct model for multilingual chat, coding, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2025-07-21",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.22,
"output": 0.8,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crusoe/Qwen/Qwen3-235B-A22B-Instruct-2507\", apiKey: processEnvironment[\"CRUSOE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.crusoecloud.com/v1\")!,\n apiKey: processEnvironment[\"CRUSOE_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.3-70B-Instruct": {
"id": "meta-llama/Llama-3.3-70B-Instruct",
"name": "Llama-3.3-70B-Instruct",
"description": "Popular open Llama workhorse for multilingual chat, coding, and self-hosting",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2023-12",
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 4096
},
"cost": {
"input": 0.25,
"output": 0.75,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crusoe/meta-llama/Llama-3.3-70B-Instruct\", apiKey: processEnvironment[\"CRUSOE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.crusoecloud.com/v1\")!,\n apiKey: processEnvironment[\"CRUSOE_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.3-70B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crusoe/openai/gpt-oss-120b\", apiKey: processEnvironment[\"CRUSOE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.crusoecloud.com/v1\")!,\n apiKey: processEnvironment[\"CRUSOE_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6": {
"id": "moonshotai/Kimi-K2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.7,
"output": 3.5,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crusoe/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"CRUSOE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.crusoecloud.com/v1\")!,\n apiKey: processEnvironment[\"CRUSOE_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/GLM-5.1": {
"id": "zai/GLM-5.1",
"name": "GLM-5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4.4,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crusoe/zai/GLM-5.1\", apiKey: processEnvironment[\"CRUSOE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.crusoecloud.com/v1\")!,\n apiKey: processEnvironment[\"CRUSOE_API_KEY\"]\n)\nlet session = provider.model(\"zai/GLM-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai/GLM-5.2": {
"id": "zai/GLM-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"crusoe/zai/GLM-5.2\", apiKey: processEnvironment[\"CRUSOE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.inference.crusoecloud.com/v1\")!,\n apiKey: processEnvironment[\"CRUSOE_API_KEY\"]\n)\nlet session = provider.model(\"zai/GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"empiriolabs": {
"id": "empiriolabs",
"name": "EmpirioLabs AI",
"baseURL": "https://api.empiriolabs.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"EMPIRIOLABS_API_KEY"
],
"doc": "https://docs.empiriolabs.ai",
"modelCount": 60,
"models": {
"glm-5-1": {
"id": "glm-5-1",
"name": "GLM 5.1",
"description": "Strong GLM coding model for agentic engineering, terminals, and repository generation",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 38912
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-07",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202000,
"output": 128000
},
"cost": {
"input": 0.825,
"output": 3.301,
"cache_read": 0.165,
"tiers": [
{
"input": 1.1,
"output": 3.851,
"cache_read": 0.22,
"tier": {
"type": "context",
"size": 32000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/glm-5-1\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"glm-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-5-9b": {
"id": "qwen3-5-9b",
"name": "Qwen3.5 9B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.09,
"output": 0.13,
"cache_read": 0.045
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-5-9b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro-0813": {
"id": "deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 393216
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 393216
},
"cost": {
"input": 1.32,
"output": 3.96,
"cache_read": 1.32
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/deepseek-v4-pro-0813\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-7-code-highspeed": {
"id": "kimi-k2-7-code-highspeed",
"name": "Kimi K2.7 Code Highspeed",
"description": "Lower-latency Kimi Code variant for interactive edits and coding-agent loops",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 131072
},
"cost": {
"input": 1.9,
"output": 8,
"cache_read": 1.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/kimi-k2-7-code-highspeed\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-7-code-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash-0731": {
"id": "deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 393216
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 393216
},
"cost": {
"input": 0.424,
"output": 1.272,
"cache_read": 0.424
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/deepseek-v4-flash-0731\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mistral-small-4": {
"id": "mistral-small-4",
"name": "Mistral Small 4",
"description": "Fast Mistral production model for chat, extraction, and cost-sensitive agents",
"family": "mistral-small",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2026-03-16",
"last_updated": "2026-03-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/mistral-small-4\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"mistral-small-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2-7-highspeed": {
"id": "minimax-m2-7-highspeed",
"name": "MiniMax M2.7 Highspeed",
"description": "Low-latency M2.7 variant for interactive coding plans and agent loops",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 32768
},
"cost": {
"input": 0.3,
"output": 1.2,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/minimax-m2-7-highspeed\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2-7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-3-27b": {
"id": "gemma-3-27b",
"name": "Gemma 3 27B",
"description": "Largest open Gemma 3 instruction model for multilingual text generation and visual understanding",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-03-12",
"last_updated": "2025-03-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/gemma-3-27b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"gemma-3-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-8-max-0902": {
"id": "qwen3-8-max-0902",
"name": "Qwen3.8 Max 0902",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 262144
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-8-max-0902\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-8-max-0902\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-2-0-pro": {
"id": "seed-2-0-pro",
"name": "Seed 2.0 Pro",
"description": "Flagship ByteDance Seed 2.0 model for complex multimodal reasoning and long-horizon agent workflows",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.63,
"output": 3.79,
"cache_read": 0.63,
"tiers": [
{
"input": 1.26,
"output": 7.58,
"cache_read": 1.26,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/seed-2-0-pro\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"seed-2-0-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-7-plus": {
"id": "qwen3-7-plus",
"name": "Qwen3.7 Plus",
"description": "Multimodal Qwen workhorse for long-context agents, visual inputs, and coding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 256000
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.4,
"output": 1.6,
"cache_read": 0.4,
"tiers": [
{
"input": 1.2,
"output": 4.8,
"cache_read": 1.2,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 1.2,
"output": 4.8,
"cache_read": 1.2
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-7-plus\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-6-flash": {
"id": "qwen3-6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 64000
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.25,
"tiers": [
{
"input": 1,
"output": 4,
"cache_read": 1,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 1,
"output": 4,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-6-flash\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-5-27b": {
"id": "qwen3-5-27b",
"name": "Qwen3.5 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 80000
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0.086,
"output": 0.688,
"cache_read": 0.086,
"tiers": [
{
"input": 0.258,
"output": 2.064,
"cache_read": 0.258,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-5-27b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-5-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4-6v-flash": {
"id": "glm-4-6v-flash",
"name": "GLM 4.6V Flash",
"description": "Lightweight GLM vision model for visual reasoning, documents, and multimodal agents",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-12-08",
"last_updated": "2025-12-08",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/glm-4-6v-flash\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"glm-4-6v-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-2-0-mini": {
"id": "seed-2-0-mini",
"name": "Seed 2.0 Mini",
"description": "Lightweight ByteDance Seed 2.0 model for low-latency multimodal reasoning and high-volume tasks",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.12,
"output": 0.5,
"cache_read": 0.12,
"tiers": [
{
"input": 0.24,
"output": 1,
"cache_read": 0.24,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/seed-2-0-mini\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"seed-2-0-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m3": {
"id": "minimax-m3",
"name": "MiniMax M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 524288
},
"cost": {
"input": 0.225,
"output": 0.9,
"cache_read": 0.045,
"tiers": [
{
"input": 0.45,
"output": 1.8,
"cache_read": 0.09,
"tier": {
"type": "context",
"size": 512000
}
}
],
"context_over_200k": {
"input": 0.45,
"output": 1.8,
"cache_read": 0.09
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/minimax-m3\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 393216
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 393216
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/deepseek-v4-flash\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-5-4b": {
"id": "qwen3-5-4b",
"name": "Qwen3.5 4B",
"description": "Qwen3.5 4B is a low-cost multimodal reasoning model with 256K context, image and video input, function tools, and structured output.",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1024,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-02",
"last_updated": "2026-03-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.04,
"output": 0.07,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-5-4b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-5-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5-3": {
"id": "glm-5-3",
"name": "GLM 5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 1.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/glm-5-3\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"glm-5-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-7-code": {
"id": "kimi-k2-7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 131072
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/kimi-k2-7-code\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fugu-ultra-v1-1": {
"id": "fugu-ultra-v1-1",
"name": "Fugu Ultra v1.1",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/fugu-ultra-v1-1\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"fugu-ultra-v1-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3-5-flash-2603": {
"id": "step-3-5-flash-2603",
"name": "Step 3.5 Flash 2603",
"description": "StepFun flash model for efficient multimodal reasoning, coding, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/step-3-5-flash-2603\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"step-3-5-flash-2603\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-5-397b-a17b": {
"id": "qwen3-5-397b-a17b",
"name": "Qwen3.5 397B-A17B",
"description": "Large open Qwen multimodal MoE for visual agents and long technical tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 80000
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0.172,
"output": 1.032,
"cache_read": 0.172,
"tiers": [
{
"input": 0.43,
"output": 2.58,
"cache_read": 0.43,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-5-397b-a17b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-2-1-turbo": {
"id": "seed-2-1-turbo",
"name": "Seed 2.1 Turbo",
"description": "Faster ByteDance Seed 2.1 model for multimodal reasoning and latency-sensitive agent workflows",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-23",
"last_updated": "2026-06-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 0.63,
"output": 3.13,
"cache_read": 0.63
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/seed-2-1-turbo\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"seed-2-1-turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v3-2": {
"id": "deepseek-v3-2",
"name": "DeepSeek V3.2",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes, sparse attention, and tool-use",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 393216
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-12-01",
"last_updated": "2025-12-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.57,
"output": 1.71,
"cache_read": 0.57
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/deepseek-v3-2\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v3-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4-7-flash": {
"id": "glm-4-7-flash",
"name": "GLM 4.7 Flash",
"description": "Budget GLM lane for fast coding help, routing, and everyday automation",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/glm-4-7-flash\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"glm-4-7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2-6": {
"id": "kimi-k2-6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 81920
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16000
},
"cost": {
"input": 0.8939,
"output": 3.7131,
"cache_read": 0.1788
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/kimi-k2-6\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-26b-a4b": {
"id": "gemma-4-26b-a4b",
"name": "Gemma 4 26B-A4B",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.29,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/gemma-4-26b-a4b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-26b-a4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-6-35b-a3b": {
"id": "qwen3-6-35b-a3b",
"name": "Qwen3.6 35B A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.07,
"output": 0.42,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-6-35b-a3b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-5-35b-a3b": {
"id": "qwen3-5-35b-a3b",
"name": "Qwen3.5 35B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 80000
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0.057,
"output": 0.459,
"cache_read": 0.057,
"tiers": [
{
"input": 0.229,
"output": 1.835,
"cache_read": 0.229,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-5-35b-a3b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-5-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1-2": {
"id": "muse-spark-1-2",
"name": "Muse Spark 1.2",
"description": "Muse Spark 1.2 is a coding-focused update to Muse Spark 1.1 with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-05",
"last_updated": "2026-08-05",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/muse-spark-1-2\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-6-plus": {
"id": "qwen3-6-plus",
"name": "Qwen3.6 Plus",
"description": "Earlier Qwen multimodal workhorse for million-token agent and document tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.5,
"tiers": [
{
"input": 2,
"output": 6,
"cache_read": 2,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 2,
"output": 6,
"cache_read": 2
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-6-plus\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k3": {
"id": "kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/kimi-k3\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-5-122b-a10b": {
"id": "qwen3-5-122b-a10b",
"name": "Qwen3.5 122B-A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 80000
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0.115,
"output": 0.917,
"cache_read": 0.115,
"tiers": [
{
"input": 0.287,
"output": 2.294,
"cache_read": 0.287,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-5-122b-a10b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-5-122b-a10b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-max": {
"id": "qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 1.08,
"output": 5.52,
"cache_read": 1.08,
"tiers": [
{
"input": 2.16,
"output": 11.04,
"cache_read": 2.16,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 2.7,
"output": 13.8,
"cache_read": 2.7,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-max\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-4-5-flash": {
"id": "glm-4-5-flash",
"name": "GLM 4.5 Flash",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 98304
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/glm-4-5-flash\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"glm-4-5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5-3-flash": {
"id": "glm-5-3-flash",
"name": "GLM 5.3 Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.075,
"output": 0.25,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/glm-5-3-flash\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"glm-5-3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3-5-flash": {
"id": "step-3-5-flash",
"name": "Step 3.5 Flash",
"description": "StepFun flash lane for quick multimodal reasoning and coding assistance",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-29",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 131072
},
"cost": {
"input": 0.1,
"output": 0.3,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/step-3-5-flash\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"step-3-5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-glimmer-30b": {
"id": "muse-glimmer-30b",
"name": "Muse Glimmer 30B",
"description": "Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-01-04",
"release_date": "2026-08-10",
"last_updated": "2026-08-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.8,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/muse-glimmer-30b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"muse-glimmer-30b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-6-27b": {
"id": "qwen3-6-27b",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 80000
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 64000
},
"cost": {
"input": 0.412564,
"output": 2.475384,
"cache_read": 0.412564
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-6-27b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-8-27b": {
"id": "qwen3-8-27b",
"name": "Qwen3.8 27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.17,
"output": 0.5,
"cache_read": 0.08
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-8-27b\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1-1": {
"id": "muse-spark-1-1",
"name": "Muse Spark 1.1",
"description": "Muse Spark is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-08",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/muse-spark-1-1\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5-2": {
"id": "glm-5-2",
"name": "GLM 5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"minimal",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 1.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/glm-5-2\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"glm-5-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-2-0-code": {
"id": "seed-2-0-code",
"name": "Seed 2.0 Code",
"description": "ByteDance Seed coding model for multimodal software engineering and long-running agents",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.4,
"output": 2.4,
"cache_read": 0.4,
"tiers": [
{
"input": 0.8,
"output": 4.8,
"cache_read": 0.8,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/seed-2-0-code\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"seed-2-0-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-7-max": {
"id": "qwen3-7-max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 64000
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-7-max\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-5": {
"id": "mimo-v2-5",
"name": "MiMo V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 0.7,
"output": 1.4,
"cache_read": 0.014
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/mimo-v2-5\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-5-flash": {
"id": "qwen3-5-flash",
"name": "Qwen3.5 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.09,
"output": 0.368,
"cache_read": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-5-flash\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"seed-2-0-lite": {
"id": "seed-2-0-lite",
"name": "Seed 2.0 Lite",
"description": "Cost-efficient ByteDance Seed 2.0 model for production chat, analysis, and structured generation",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.31,
"output": 2.5,
"cache_read": 0.31,
"tiers": [
{
"input": 0.62,
"output": 5,
"cache_read": 0.62,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/seed-2-0-lite\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"seed-2-0-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"muse-spark-1-3": {
"id": "muse-spark-1-3",
"name": "Muse Spark 1.3",
"description": "Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It improves long-horizon agent collaboration, instruction following, and coding efficiency relative to Muse Spark 1.2.",
"family": "muse",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 4.25,
"cache_read": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/muse-spark-1-3\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"muse-spark-1-3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v2-5-pro": {
"id": "mimo-v2-5-pro",
"name": "MiMo V2.5 Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2.175,
"output": 4.35,
"cache_read": 0.018
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/mimo-v2-5-pro\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v2-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"step-3-7-flash": {
"id": "step-3-7-flash",
"name": "Step 3.7 Flash",
"description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"input": 256000,
"output": 131072
},
"cost": {
"input": 0.2,
"output": 1.15,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/step-3-7-flash\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"step-3-7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-pro": {
"id": "deepseek-v4-pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 393216
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 393216
},
"cost": {
"input": 1.65,
"output": 3.3,
"cache_read": 1.65
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/deepseek-v4-pro\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fugu-ultra-v1-0": {
"id": "fugu-ultra-v1-0",
"name": "Fugu Ultra v1.0",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 7.5,
"output": 45,
"cache_read": 1.5,
"tiers": [
{
"input": 15,
"output": 67.5,
"cache_read": 3,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 15,
"output": 67.5,
"cache_read": 3
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/fugu-ultra-v1-0\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"fugu-ultra-v1-0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-8-max": {
"id": "qwen3-8-max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 262144
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-8-max\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-8-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"minimax-m2-7": {
"id": "minimax-m2-7",
"name": "MiniMax M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 200000,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 0.6,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/minimax-m2-7\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"minimax-m2-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-5-plus": {
"id": "qwen3-5-plus",
"name": "Qwen3.5 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.36,
"output": 2.21,
"cache_read": 0.36,
"tiers": [
{
"input": 1.08,
"output": 6.62,
"cache_read": 1.08,
"tier": {
"type": "context",
"size": 256000
}
}
],
"context_over_200k": {
"input": 1.08,
"output": 6.62,
"cache_read": 1.08
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-5-plus\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-5-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-7-flash": {
"id": "qwen3-7-flash",
"name": "Qwen3.7 Flash",
"description": "Lightweight multimodal Qwen model for high-throughput text, image, and video tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 131072
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 991000,
"output": 65536
},
"cost": {
"input": 0.03,
"output": 0.13,
"cache_read": 0.006,
"tiers": [
{
"input": 0.1,
"output": 0.4,
"cache_read": 0.02,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 0.2,
"output": 0.8,
"cache_read": 0.04,
"tier": {
"type": "context",
"size": 256000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-7-flash\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-8-flash": {
"id": "qwen3-8-flash",
"name": "Qwen3.8 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 262144
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.16,
"output": 0.47,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-8-flash\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-6-max-preview": {
"id": "qwen3-6-max-preview",
"name": "Qwen3.6 Max Preview",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
},
{
"type": "budget_tokens",
"min": 1,
"max": 393216
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-20",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 1.31,
"output": 7.88,
"cache_read": 1.31,
"tiers": [
{
"input": 1.97,
"output": 11.82,
"cache_read": 1.97,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/qwen3-6-max-preview\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-6-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fugu-ultra-v2-0": {
"id": "fugu-ultra-v2-0",
"name": "Fugu Ultra v2.0",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"empiriolabs/fugu-ultra-v2-0\", apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.empiriolabs.ai/v1\")!,\n apiKey: processEnvironment[\"EMPIRIOLABS_API_KEY\"]\n)\nlet session = provider.model(\"fugu-ultra-v2-0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"klokintegration": {
"id": "klokintegration",
"name": "klokintegration.se",
"baseURL": "https://api-gw.klok.ipaas.se/proxy/kloker-key/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"KLOKINTEGRATION_API_KEY"
],
"doc": "https://klokintegration.se/docs/ai-api",
"modelCount": 3,
"models": {
"Kloker-Integration-Developer": {
"id": "Kloker-Integration-Developer",
"name": "Kloker Integration Developer",
"description": "Knows the customer integration environment and Klok best practices. Opinionated about implementation. The gateway runs ecosystem lookup tools server-side and appends a system-prompt injection. Client system prompts and OpenAI tool calls are preserved.",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-29",
"last_updated": "2026-08-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 50000
},
"status": "beta",
"cost": {
"input": 0.23,
"output": 1.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"klokintegration/Kloker-Integration-Developer\", apiKey: processEnvironment[\"KLOKINTEGRATION_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-gw.klok.ipaas.se/proxy/kloker-key/v1\")!,\n apiKey: processEnvironment[\"KLOKINTEGRATION_API_KEY\"]\n)\nlet session = provider.model(\"Kloker-Integration-Developer\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Kloker-Integration-Architect": {
"id": "Kloker-Integration-Architect",
"name": "Kloker Integration Architect",
"description": "Knows the customer integration environment and Klok best practices. Opinionated about structure. The gateway runs ecosystem lookup tools server-side and appends a system-prompt injection (data contracts, CloudEvents, event-driven flows). Client system prompts and OpenAI tool calls are preserved.",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-29",
"last_updated": "2026-08-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 50000
},
"status": "beta",
"cost": {
"input": 0.23,
"output": 1.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"klokintegration/Kloker-Integration-Architect\", apiKey: processEnvironment[\"KLOKINTEGRATION_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-gw.klok.ipaas.se/proxy/kloker-key/v1\")!,\n apiKey: processEnvironment[\"KLOKINTEGRATION_API_KEY\"]\n)\nlet session = provider.model(\"Kloker-Integration-Architect\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Kloker": {
"id": "Kloker",
"name": "Kloker",
"description": "Cheap general model with a clean context. Nothing from the customer environment is packed in. It tracks the current best open source model. The Klok team verifies it and upgrades it periodically.",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-08-29",
"last_updated": "2026-08-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 50000
},
"status": "beta",
"cost": {
"input": 0.23,
"output": 1.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"klokintegration/Kloker\", apiKey: processEnvironment[\"KLOKINTEGRATION_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api-gw.klok.ipaas.se/proxy/kloker-key/v1\")!,\n apiKey: processEnvironment[\"KLOKINTEGRATION_API_KEY\"]\n)\nlet session = provider.model(\"Kloker\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"privatemode-ai": {
"id": "privatemode-ai",
"name": "Privatemode AI",
"baseURL": "http://localhost:8080/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"PRIVATEMODE_API_KEY",
"PRIVATEMODE_ENDPOINT"
],
"doc": "https://docs.privatemode.ai/api/overview",
"modelCount": 9,
"models": {
"kimi-latest": {
"id": "kimi-latest",
"name": "Kimi (latest)",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 262144
},
"cost": {
"input": 1.791,
"output": 8.9436,
"cache_read": 0.1733
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"privatemode-ai/kimi-latest\", apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://localhost:8080/v1\")!,\n apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"kimi-k2.6": {
"id": "kimi-k2.6",
"name": "Kimi K2.6",
"description": "Multimodal Kimi workhorse for agent loops, coding tasks, and visual context",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 262144
},
"cost": {
"input": 1.791,
"output": 8.9436,
"cache_read": 0.1733
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"privatemode-ai/kimi-k2.6\", apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://localhost:8080/v1\")!,\n apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"]\n)\nlet session = provider.model(\"kimi-k2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"voxtral-mini-3b": {
"id": "voxtral-mini-3b",
"name": "Voxtral Mini 3B",
"description": "Speech-to-text model for audio transcription, translation, and audio understanding",
"family": "voxtral",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-07",
"last_updated": "2025-07",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 32000
},
"cost": {
"input": 0.00462,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"privatemode-ai/voxtral-mini-3b\", apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://localhost:8080/v1\")!,\n apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"]\n)\nlet session = provider.model(\"voxtral-mini-3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen3-embedding-4b": {
"id": "qwen3-embedding-4b",
"name": "Qwen3-Embedding 4B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"knowledge": "2025-06",
"release_date": "2025-06-06",
"last_updated": "2025-06-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32000,
"output": 2560
},
"cost": {
"input": 0.1502,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"privatemode-ai/qwen3-embedding-4b\", apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://localhost:8080/v1\")!,\n apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"]\n)\nlet session = provider.model(\"qwen3-embedding-4b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"whisper-large-v3": {
"id": "whisper-large-v3",
"name": "Whisper large-v3",
"description": "Open Whisper checkpoint for robust multilingual transcription and captioning",
"family": "whisper",
"attachment": true,
"reasoning": false,
"tool_call": false,
"release_date": "2024-10-01",
"last_updated": "2024-10-01",
"modalities": {
"input": [
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 448,
"output": 4096
},
"cost": {
"input": 0.01618,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"privatemode-ai/whisper-large-v3\", apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://localhost:8080/v1\")!,\n apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"]\n)\nlet session = provider.model(\"whisper-large-v3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-latest": {
"id": "glm-latest",
"name": "GLM (latest)",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 131072
},
"status": "beta",
"cost": {
"input": 1.791,
"output": 8.9436,
"cache_read": 0.1733
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"privatemode-ai/glm-latest\", apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://localhost:8080/v1\")!,\n apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"]\n)\nlet session = provider.model(\"glm-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gpt-oss-120b": {
"id": "gpt-oss-120b",
"name": "gpt-oss-120b",
"description": "Open GPT reasoning model for self-hosted agents and controllable deployments",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 32768
},
"cost": {
"input": 0.4969,
"output": 1.9644,
"cache_read": 0.0462
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"privatemode-ai/gpt-oss-120b\", apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://localhost:8080/v1\")!,\n apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"]\n)\nlet session = provider.model(\"gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ocr-2": {
"id": "deepseek-ocr-2",
"name": "DeepSeek OCR 2",
"description": "High-accuracy OCR model for extracting text from documents, screenshots, receipts, and natural scenes",
"attachment": true,
"reasoning": false,
"tool_call": false,
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 8192,
"output": 8192
},
"status": "beta",
"cost": {
"input": 0.8897,
"output": 1.4675,
"cache_read": 0.0924
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"privatemode-ai/deepseek-ocr-2\", apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://localhost:8080/v1\")!,\n apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ocr-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm-5.3": {
"id": "glm-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 131072
},
"status": "beta",
"cost": {
"input": 1.791,
"output": 8.9436,
"cache_read": 0.1733
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"privatemode-ai/glm-5.3\", apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://localhost:8080/v1\")!,\n apiKey: processEnvironment[\"PRIVATEMODE_API_KEY\"]\n)\nlet session = provider.model(\"glm-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"minimax-coding-plan": {
"id": "minimax-coding-plan",
"name": "MiniMax Token Plan (minimax.io)",
"baseURL": "https://api.minimax.io/anthropic/v1",
"npm": "@ai-sdk/anthropic",
"swiftDriver": "anthropicMessages",
"env": [
"MINIMAX_API_KEY"
],
"doc": "https://platform.minimax.io/docs/token-plan/intro",
"modelCount": 7,
"models": {
"MiniMax-M2": {
"id": "MiniMax-M2",
"name": "MiniMax-M2",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-10-27",
"last_updated": "2025-10-27",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-coding-plan/MiniMax-M2\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.1": {
"id": "MiniMax-M2.1",
"name": "MiniMax-M2.1",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-23",
"last_updated": "2025-12-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-coding-plan/MiniMax-M2.1\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.5": {
"id": "MiniMax-M2.5",
"name": "MiniMax-M2.5",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-coding-plan/MiniMax-M2.5\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.5-highspeed": {
"id": "MiniMax-M2.5-highspeed",
"name": "MiniMax-M2.5-highspeed",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-02-13",
"last_updated": "2026-02-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-coding-plan/MiniMax-M2.5-highspeed\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.5-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M3": {
"id": "MiniMax-M3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal coding model for long-context reasoning and agent tasks",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-25",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 512000
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-coding-plan/MiniMax-M3\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.7-highspeed": {
"id": "MiniMax-M2.7-highspeed",
"name": "MiniMax-M2.7-highspeed",
"description": "High-speed MiniMax model for low-latency coding and agent workflows",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-coding-plan/MiniMax-M2.7-highspeed\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.7-highspeed\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M2.7": {
"id": "MiniMax-M2.7",
"name": "MiniMax-M2.7",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 204800,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "anthropicMessages",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"minimax-coding-plan/MiniMax-M2.7\", apiKey: processEnvironment[\"MINIMAX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.minimax.io/anthropic/v1\")!,\n apiKey: processEnvironment[\"MINIMAX_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"inferx": {
"id": "inferx",
"name": "InferX",
"baseURL": "https://model.inferx.net/endpoints/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"INFERX_API_KEY"
],
"doc": "https://model.inferx.net/endpoints",
"modelCount": 12,
"models": {
"gemma-4-31B-it-fp8": {
"id": "gemma-4-31B-it-fp8",
"name": "Gemma 4 31B IT FP8",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/gemma-4-31B-it-fp8\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-31B-it-fp8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3-Coder-Next-FP8-no-thinking": {
"id": "Qwen3-Coder-Next-FP8-no-thinking",
"name": "Qwen3-Coder-Next-FP8-no-thinking",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-03",
"last_updated": "2026-02-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 260000,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/Qwen3-Coder-Next-FP8-no-thinking\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3-Coder-Next-FP8-no-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-v4-flash": {
"id": "deepseek-v4-flash",
"name": "deepseek-v4-flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 100000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/deepseek-v4-flash\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-v4-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Devstral-2-123B-Instruct-2512-int4-AutoRound": {
"id": "Devstral-2-123B-Instruct-2512-int4-AutoRound",
"name": "Devstral-2-123B-Instruct-2512-int4-AutoRound",
"description": "Mistral's coding-agent model for repository work, terminal tasks, and software fixes",
"family": "devstral",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2025-12-09",
"last_updated": "2025-12-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 128000,
"output": 128000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/Devstral-2-123B-Instruct-2512-int4-AutoRound\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"Devstral-2-123B-Instruct-2512-int4-AutoRound\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Agents-A1": {
"id": "Agents-A1",
"name": "Agents-A1",
"description": "35B MoE agentic model built for long-horizon search, engineering, and scientific reasoning tasks",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"release_date": "2026-06-26",
"last_updated": "2026-06-26",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 100000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/Agents-A1\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"Agents-A1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Ornith-1.0-35B-FP8": {
"id": "Ornith-1.0-35B-FP8",
"name": "Ornith-1.0-35B-FP8",
"description": "Large coding-reasoning model for agentic software tasks and RL search",
"family": "ornith",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-06-25",
"last_updated": "2026-06-25",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 100000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/Ornith-1.0-35B-FP8\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"Ornith-1.0-35B-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3.6-35B-A3B-FP8": {
"id": "Qwen3.6-35B-A3B-FP8",
"name": "Qwen3.6 35B A3B FP8",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/Qwen3.6-35B-A3B-FP8\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3.6-35B-A3B-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3.6-27B-FP8": {
"id": "Qwen3.6-27B-FP8",
"name": "Qwen3.6 27B FP8",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/Qwen3.6-27B-FP8\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3.6-27B-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3.6-35B-A3B-fp8-no-thinking": {
"id": "Qwen3.6-35B-A3B-fp8-no-thinking",
"name": "Qwen3.6-35B-A3B-fp8-no-thinking",
"description": "Qwen3.6-35B-A3B-fp8 disable thinking",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262000,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/Qwen3.6-35B-A3B-fp8-no-thinking\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3.6-35B-A3B-fp8-no-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3-Coder-Next-FP8": {
"id": "Qwen3-Coder-Next-FP8",
"name": "Qwen3 Coder Next FP8",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-02-03",
"last_updated": "2026-02-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256144,
"output": 65536
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/Qwen3-Coder-Next-FP8\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3-Coder-Next-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3-Embedding-8B": {
"id": "Qwen3-Embedding-8B",
"name": "Qwen3-Embedding-8B",
"description": "Embedding model for semantic search, retrieval, clustering, and ranking pipelines",
"family": "text-embedding",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": false,
"release_date": "2025-06-05",
"last_updated": "2025-06-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 0
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/Qwen3-Embedding-8B\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3-Embedding-8B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"mimo-v25": {
"id": "mimo-v25",
"name": "mimo-v25",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 100000
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"inferx/mimo-v25\", apiKey: processEnvironment[\"INFERX_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://model.inferx.net/endpoints/v1\")!,\n apiKey: processEnvironment[\"INFERX_API_KEY\"]\n)\nlet session = provider.model(\"mimo-v25\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"umans-ai-coding-plan": {
"id": "umans-ai-coding-plan",
"name": "Umans AI Coding Plan",
"baseURL": "https://api.code.umans.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"UMANS_AI_CODING_PLAN_API_KEY"
],
"doc": "https://app.umans.ai/offers/code/docs",
"modelCount": 8,
"models": {
"umans-qwen3.6-35b-a3b": {
"id": "umans-qwen3.6-35b-a3b",
"name": "Qwen3.6 35B A3B",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai-coding-plan/umans-qwen3.6-35b-a3b\", apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"umans-qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-glm-5.2": {
"id": "umans-glm-5.2",
"name": "GLM 5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 405504,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai-coding-plan/umans-glm-5.2\", apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"umans-glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-kimi-k3": {
"id": "umans-kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai-coding-plan/umans-kimi-k3\", apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"umans-kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-kimi-k2.7": {
"id": "umans-kimi-k2.7",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai-coding-plan/umans-kimi-k2.7\", apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"umans-kimi-k2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-deepseek-v4-flash-0731": {
"id": "umans-deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393215
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai-coding-plan/umans-deepseek-v4-flash-0731\", apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"umans-deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-coder": {
"id": "umans-coder",
"name": "Umans Coder",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai-coding-plan/umans-coder\", apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"umans-coder\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-flash": {
"id": "umans-flash",
"name": "Umans Flash",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai-coding-plan/umans-flash\", apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"umans-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-deepseek-v4-pro-0813": {
"id": "umans-deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393215
},
"cost": {
"input": 0,
"output": 0,
"cache_read": 0,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai-coding-plan/umans-deepseek-v4-pro-0813\", apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_CODING_PLAN_API_KEY\"]\n)\nlet session = provider.model(\"umans-deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"databricks": {
"id": "databricks",
"name": "Databricks",
"baseURL": "https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"DATABRICKS_HOST",
"DATABRICKS_TOKEN"
],
"doc": "https://docs.databricks.com/aws/en/machine-learning/foundation-models/",
"modelCount": 30,
"models": {
"databricks-claude-opus-4-5": {
"id": "databricks-claude-opus-4-5",
"name": "Claude Opus 4.5 (latest)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2025-11-24",
"last_updated": "2025-11-24",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-claude-opus-4-5\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-claude-opus-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-claude-sonnet-4-6": {
"id": "databricks-claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"description": "Claude workhorse for coding agents, careful analysis, and production cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-02-17",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-claude-sonnet-4-6\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-claude-sonnet-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gemini-3-pro": {
"id": "databricks-gemini-3-pro",
"name": "Gemini 3 Pro Preview",
"description": "Preview Gemini flagship for complex reasoning, coding, and rich multimodal prompts",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-11-18",
"last_updated": "2025-11-18",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gemini-3-pro\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gemini-3-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-kimi-k2-7-code": {
"id": "databricks-kimi-k2-7-code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-kimi-k2-7-code\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-kimi-k2-7-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5-6-luna": {
"id": "databricks-gpt-5-6-luna",
"name": "GPT-5.6 Luna",
"description": "Cost-efficient GPT-5.6 model for fast, high-volume workloads",
"family": "gpt-luna",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1,
"output": 6,
"cache_read": 0.1,
"tiers": [
{
"input": 2,
"output": 9,
"cache_read": 0.2,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 2,
"output": 9,
"cache_read": 0.2
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5-6-luna\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5-6-luna\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-claude-opus-4-1": {
"id": "databricks-claude-opus-4-1",
"name": "Claude Opus 4.1 (latest)",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 32000
},
"cost": {
"input": 15,
"output": 75,
"cache_read": 1.5,
"cache_write": 18.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-claude-opus-4-1\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-claude-opus-4-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5-mini": {
"id": "databricks-gpt-5-mini",
"name": "GPT-5 Mini",
"description": "Small GPT-5 for responsive agents, coding help, and everyday automation",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.25,
"output": 2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5-mini\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gemini-2-5-flash": {
"id": "databricks-gemini-2-5-flash",
"name": "Gemini 2.5 Flash",
"description": "Fast Gemini workhorse for multimodal apps where latency and price matter",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 0,
"max": 24576
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 2.5,
"cache_read": 0.03,
"input_audio": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gemini-2-5-flash\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gemini-2-5-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-claude-haiku-4-5": {
"id": "databricks-claude-haiku-4-5",
"name": "Claude Haiku 4.5 (latest)",
"description": "Fast Claude lane for lightweight agents, office tasks, and responsive chat",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2025-02-28",
"release_date": "2025-10-15",
"last_updated": "2025-10-15",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-claude-haiku-4-5\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-claude-haiku-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-claude-sonnet-4-5": {
"id": "databricks-claude-sonnet-4-5",
"name": "Claude Sonnet 4.5 (latest)",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-claude-sonnet-4-5\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-claude-sonnet-4-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5-4": {
"id": "databricks-gpt-5-4",
"name": "GPT-5.4",
"description": "Agent-ready GPT for coding and computer-use workflows at a lower cost",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-05",
"last_updated": "2026-03-05",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5
},
"provider": {
"body": {
"service_tier": "priority"
}
}
}
}
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5-4\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5-6-sol": {
"id": "databricks-gpt-5-6-sol",
"name": "GPT-5.6 Sol",
"description": "Frontier GPT-5.6 model for complex professional work, coding, and agentic workflows",
"family": "gpt-sol",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5-6-sol\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5-6-sol\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-glm-5-2": {
"id": "databricks-glm-5-2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-glm-5-2\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-glm-5-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5-4-nano": {
"id": "databricks-gpt-5-4-nano",
"name": "GPT-5.4 nano",
"description": "Cheapest GPT-5.4 lane for simple routing, extraction, and bulk automation",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.2,
"output": 1.25,
"cache_read": 0.02
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5-4-nano\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5-4-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5-5": {
"id": "databricks-gpt-5-5",
"name": "GPT-5.5",
"description": "Default frontier GPT for coding, computer use, research, and knowledge work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-12-01",
"release_date": "2026-04-23",
"last_updated": "2026-04-23",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 12.5,
"output": 75,
"cache_read": 1.25
},
"provider": {
"body": {
"service_tier": "priority"
}
}
}
}
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5-5\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gemini-3-1-flash-lite": {
"id": "databricks-gemini-3-1-flash-lite",
"name": "Gemini 3.1 Flash Lite Preview",
"description": "Low-latency Gemini model for high-volume multimodal and agent workloads",
"family": "gemini-flash-lite",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-03-03",
"last_updated": "2026-03-03",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.25,
"output": 1.5,
"cache_read": 0.025,
"input_audio": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gemini-3-1-flash-lite\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gemini-3-1-flash-lite\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gemini-3-flash": {
"id": "databricks-gemini-3-flash",
"name": "Gemini 3 Flash Preview",
"description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
"family": "gemini-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-12-17",
"last_updated": "2025-12-17",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.05,
"input_audio": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gemini-3-flash\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gemini-3-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-claude-opus-4-7": {
"id": "databricks-claude-opus-4-7",
"name": "Claude Opus 4.7",
"description": "Stronger Opus tier for advanced software work and high-stakes reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": false,
"knowledge": "2026-01-31",
"release_date": "2026-04-16",
"last_updated": "2026-04-16",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 30,
"output": 150,
"cache_read": 3,
"cache_write": 37.5
},
"provider": {
"body": {
"speed": "fast"
},
"headers": {
"anthropic-beta": "fast-mode-2026-02-01"
}
}
}
}
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-claude-opus-4-7\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-claude-opus-4-7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-claude-opus-4-6": {
"id": "databricks-claude-opus-4-6",
"name": "Claude Opus 4.6",
"description": "High-end Claude for difficult coding, planning, and slower expert reasoning",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-05-31",
"release_date": "2026-02-05",
"last_updated": "2026-03-13",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 30,
"output": 150,
"cache_read": 3,
"cache_write": 37.5
},
"provider": {
"body": {
"speed": "fast"
},
"headers": {
"anthropic-beta": "fast-mode-2026-02-01"
}
}
}
}
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-claude-opus-4-6\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-claude-opus-4-6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-claude-sonnet-4": {
"id": "databricks-claude-sonnet-4",
"name": "Claude Sonnet 4.5",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 1024
}
],
"tool_call": true,
"temperature": true,
"knowledge": "2025-07-31",
"release_date": "2025-09-29",
"last_updated": "2025-09-29",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3,
"cache_write": 3.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-claude-sonnet-4\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-claude-sonnet-4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5-1": {
"id": "databricks-gpt-5-1",
"name": "GPT-5.1",
"description": "Sharper GPT-5 generation for coding, product work, and tool-assisted tasks",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-09-30",
"release_date": "2025-11-13",
"last_updated": "2025-11-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5-1\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5-1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-oss-20b": {
"id": "databricks-gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.05,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-oss-20b\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5-4-mini": {
"id": "databricks-gpt-5-4-mini",
"name": "GPT-5.4 mini",
"description": "Strong small GPT for coding subagents, quick tool use, and high-volume work",
"family": "gpt-mini",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2026-03-17",
"last_updated": "2026-03-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"experimental": {
"modes": {
"fast": {
"cost": {
"input": 1.5,
"output": 9,
"cache_read": 0.15
},
"provider": {
"body": {
"service_tier": "priority"
}
}
}
}
},
"cost": {
"input": 0.75,
"output": 4.5,
"cache_read": 0.075
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5-4-mini\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5-4-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gemini-3-1-pro": {
"id": "databricks-gemini-3-1-pro",
"name": "Gemini 3.1 Pro Preview Custom Tools",
"description": "Advanced Gemini model for complex reasoning, coding, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-19",
"last_updated": "2026-02-19",
"modalities": {
"input": [
"text",
"image",
"video",
"audio",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 2,
"output": 12,
"cache_read": 0.2,
"tiers": [
{
"input": 4,
"output": 18,
"cache_read": 0.4,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 4,
"output": 18,
"cache_read": 0.4
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gemini-3-1-pro\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gemini-3-1-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5-6-terra": {
"id": "databricks-gpt-5-6-terra",
"name": "GPT-5.6 Terra",
"description": "Balanced GPT-5.6 model for capable, cost-efficient everyday work",
"family": "gpt-terra",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2026-02-16",
"release_date": "2026-07-09",
"last_updated": "2026-07-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1050000,
"input": 922000,
"output": 128000
},
"cost": {
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tiers": [
{
"input": 5,
"output": 22.5,
"cache_read": 0.5,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 5,
"output": 22.5,
"cache_read": 0.5
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5-6-terra\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5-6-terra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gemini-2-5-pro": {
"id": "databricks-gemini-2-5-pro",
"name": "Gemini 2.5 Pro",
"description": "Google's proven reasoning model for coding, math, and multimodal analysis",
"family": "gemini-pro",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "budget_tokens",
"min": 128,
"max": 32768
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2025-06-17",
"last_updated": "2025-06-17",
"modalities": {
"input": [
"text",
"image",
"audio",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 65536
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125,
"tiers": [
{
"input": 2.5,
"output": 15,
"cache_read": 0.25,
"tier": {
"type": "context",
"size": 200000
}
}
],
"context_over_200k": {
"input": 2.5,
"output": 15,
"cache_read": 0.25
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gemini-2-5-pro\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gemini-2-5-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5": {
"id": "databricks-gpt-5",
"name": "GPT-5",
"description": "Original GPT-5 workhorse for reasoning, coding, writing, and tool workflows",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-09-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.25,
"output": 10,
"cache_read": 0.125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5-nano": {
"id": "databricks-gpt-5-nano",
"name": "GPT-5 Nano",
"description": "Tiny GPT-5 lane for routing, extraction, classification, and bulk jobs",
"family": "gpt-nano",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2024-05-30",
"release_date": "2025-08-07",
"last_updated": "2025-08-07",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 0.05,
"output": 0.4,
"cache_read": 0.005
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5-nano\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5-nano\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-oss-120b": {
"id": "databricks-gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.072,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-oss-120b\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"databricks-gpt-5-2": {
"id": "databricks-gpt-5-2",
"name": "GPT-5.2",
"description": "Reliable GPT generation for broad coding, writing, and tool-assisted product work",
"family": "gpt",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-08-31",
"release_date": "2025-12-11",
"last_updated": "2025-12-11",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 400000,
"input": 272000,
"output": 128000
},
"cost": {
"input": 1.75,
"output": 14,
"cache_read": 0.175
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"databricks/databricks-gpt-5-2\", apiKey: processEnvironment[\"DATABRICKS_HOST\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://${DATABRICKS_HOST}/ai-gateway/mlflow/v1\")!,\n apiKey: processEnvironment[\"DATABRICKS_HOST\"]\n)\nlet session = provider.model(\"databricks-gpt-5-2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"modal": {
"id": "modal",
"name": "Modal",
"baseURL": "https://inference.us-west.modal.direct/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"MODAL_PROXY_TOKEN"
],
"doc": "https://modal.com/docs/guide/endpoints",
"modelCount": 4,
"models": {
"zai-org/GLM-5.3-Flash": {
"id": "zai-org/GLM-5.3-Flash",
"name": "GLM 5.3 Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.45,
"output": 1.5,
"cache_read": 0.09
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modal/zai-org/GLM-5.3-Flash\", apiKey: processEnvironment[\"MODAL_PROXY_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.us-west.modal.direct/v1\")!,\n apiKey: processEnvironment[\"MODAL_PROXY_TOKEN\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.3-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/Inkling-NVFP4": {
"id": "thinkingmachines/Inkling-NVFP4",
"name": "Inkling",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": false,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 262144
},
"cost": {
"input": 1.2,
"output": 5,
"cache_read": 0.27
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modal/thinkingmachines/Inkling-NVFP4\", apiKey: processEnvironment[\"MODAL_PROXY_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.us-west.modal.direct/v1\")!,\n apiKey: processEnvironment[\"MODAL_PROXY_TOKEN\"]\n)\nlet session = provider.model(\"thinkingmachines/Inkling-NVFP4\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.8-2.4T-A95B": {
"id": "Qwen/Qwen3.8-2.4T-A95B",
"name": "Qwen3.8-Max",
"description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1010000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modal/Qwen/Qwen3.8-2.4T-A95B\", apiKey: processEnvironment[\"MODAL_PROXY_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.us-west.modal.direct/v1\")!,\n apiKey: processEnvironment[\"MODAL_PROXY_TOKEN\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.8-2.4T-A95B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K3": {
"id": "moonshotai/Kimi-K3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"reasoning": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"modal/moonshotai/Kimi-K3\", apiKey: processEnvironment[\"MODAL_PROXY_TOKEN\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://inference.us-west.modal.direct/v1\")!,\n apiKey: processEnvironment[\"MODAL_PROXY_TOKEN\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"lucidquery": {
"id": "lucidquery",
"name": "LucidQuery",
"baseURL": "https://api.lucidquery.com/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"LUCIDQUERY_API_KEY"
],
"doc": "https://lucidquery.com/docs",
"modelCount": 4,
"models": {
"lucidquery-nexus-coder": {
"id": "lucidquery-nexus-coder",
"name": "LucidQuery Nexus Coder",
"description": "Coding model for repository understanding, refactors, and agentic engineering tasks",
"family": "lucid",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"knowledge": "2025-08-01",
"release_date": "2025-09-01",
"last_updated": "2025-09-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 250000,
"output": 60000
},
"cost": {
"input": 2,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"lucidquery/lucidquery-nexus-coder\", apiKey: processEnvironment[\"LUCIDQUERY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.lucidquery.com/v1\")!,\n apiKey: processEnvironment[\"LUCIDQUERY_API_KEY\"]\n)\nlet session = provider.model(\"lucidquery-nexus-coder\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"lucidquery-agi-01-frontier": {
"id": "lucidquery-agi-01-frontier",
"name": "AGI-01 Frontier",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "agi",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2026-06-05",
"release_date": "2026-06-16",
"last_updated": "2026-06-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 300000,
"output": 120000
},
"cost": {
"input": 4.5,
"output": 22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"lucidquery/lucidquery-agi-01-frontier\", apiKey: processEnvironment[\"LUCIDQUERY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.lucidquery.com/v1\")!,\n apiKey: processEnvironment[\"LUCIDQUERY_API_KEY\"]\n)\nlet session = provider.model(\"lucidquery-agi-01-frontier\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"lucidquery-agi-01-swift": {
"id": "lucidquery-agi-01-swift",
"name": "AGI-01 Swift",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "agi",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2026-06-05",
"release_date": "2026-06-16",
"last_updated": "2026-06-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 300000,
"output": 120000
},
"cost": {
"input": 2.5,
"output": 15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"lucidquery/lucidquery-agi-01-swift\", apiKey: processEnvironment[\"LUCIDQUERY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.lucidquery.com/v1\")!,\n apiKey: processEnvironment[\"LUCIDQUERY_API_KEY\"]\n)\nlet session = provider.model(\"lucidquery-agi-01-swift\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"lucidnova-rf1-100b": {
"id": "lucidnova-rf1-100b",
"name": "LucidNova RF1 100B",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"family": "nova",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": false,
"knowledge": "2025-09-16",
"release_date": "2024-12-28",
"last_updated": "2025-09-10",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 120000,
"output": 8000
},
"cost": {
"input": 2,
"output": 5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"lucidquery/lucidnova-rf1-100b\", apiKey: processEnvironment[\"LUCIDQUERY_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.lucidquery.com/v1\")!,\n apiKey: processEnvironment[\"LUCIDQUERY_API_KEY\"]\n)\nlet session = provider.model(\"lucidnova-rf1-100b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"atomic-chat": {
"id": "atomic-chat",
"name": "Atomic Chat",
"baseURL": "http://127.0.0.1:1337/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"ATOMIC_CHAT_API_KEY"
],
"doc": "https://atomic.chat",
"modelCount": 5,
"models": {
"Meta-Llama-3_1-8B-Instruct-GGUF": {
"id": "Meta-Llama-3_1-8B-Instruct-GGUF",
"name": "Meta Llama 3.1 8B Instruct (GGUF)",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2024-07-23",
"last_updated": "2024-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 4096
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"atomic-chat/Meta-Llama-3_1-8B-Instruct-GGUF\", apiKey: processEnvironment[\"ATOMIC_CHAT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://127.0.0.1:1337/v1\")!,\n apiKey: processEnvironment[\"ATOMIC_CHAT_API_KEY\"]\n)\nlet session = provider.model(\"Meta-Llama-3_1-8B-Instruct-GGUF\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3_5-9B-Q4_K_M": {
"id": "Qwen3_5-9B-Q4_K_M",
"name": "Qwen 3.5 9B (Q4_K_M)",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2026-03-05",
"last_updated": "2026-04-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"atomic-chat/Qwen3_5-9B-Q4_K_M\", apiKey: processEnvironment[\"ATOMIC_CHAT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://127.0.0.1:1337/v1\")!,\n apiKey: processEnvironment[\"ATOMIC_CHAT_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3_5-9B-Q4_K_M\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen3_5-9B-MLX-4bit": {
"id": "Qwen3_5-9B-MLX-4bit",
"name": "Qwen 3.5 9B (MLX 4-bit)",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"temperature": true,
"release_date": "2026-03-05",
"last_updated": "2026-04-04",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"atomic-chat/Qwen3_5-9B-MLX-4bit\", apiKey: processEnvironment[\"ATOMIC_CHAT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://127.0.0.1:1337/v1\")!,\n apiKey: processEnvironment[\"ATOMIC_CHAT_API_KEY\"]\n)\nlet session = provider.model(\"Qwen3_5-9B-MLX-4bit\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-E4B-it-MLX-4bit": {
"id": "gemma-4-E4B-it-MLX-4bit",
"name": "Gemma 4 E4B Instruct (MLX 4-bit)",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"atomic-chat/gemma-4-E4B-it-MLX-4bit\", apiKey: processEnvironment[\"ATOMIC_CHAT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://127.0.0.1:1337/v1\")!,\n apiKey: processEnvironment[\"ATOMIC_CHAT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-E4B-it-MLX-4bit\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"gemma-4-E4B-it-IQ4_XS": {
"id": "gemma-4-E4B-it-IQ4_XS",
"name": "Gemma 4 E4B Instruct (IQ4_XS)",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 32768,
"output": 8192
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"atomic-chat/gemma-4-E4B-it-IQ4_XS\", apiKey: processEnvironment[\"ATOMIC_CHAT_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"http://127.0.0.1:1337/v1\")!,\n apiKey: processEnvironment[\"ATOMIC_CHAT_API_KEY\"]\n)\nlet session = provider.model(\"gemma-4-E4B-it-IQ4_XS\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"umans-ai": {
"id": "umans-ai",
"name": "Umans AI",
"baseURL": "https://api.code.umans.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"UMANS_AI_API_KEY"
],
"doc": "https://app.umans.ai/offers/code/docs/orgs",
"modelCount": 7,
"models": {
"umans-glm-5.2": {
"id": "umans-glm-5.2",
"name": "GLM 5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 405504,
"output": 131072
},
"cost": {
"input": 1.4,
"output": 4.4,
"cache_read": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai/umans-glm-5.2\", apiKey: processEnvironment[\"UMANS_AI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_API_KEY\"]\n)\nlet session = provider.model(\"umans-glm-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-kimi-k3": {
"id": "umans-kimi-k3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 15,
"cache_read": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai/umans-kimi-k3\", apiKey: processEnvironment[\"UMANS_AI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_API_KEY\"]\n)\nlet session = provider.model(\"umans-kimi-k3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-kimi-k2.7": {
"id": "umans-kimi-k2.7",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai/umans-kimi-k2.7\", apiKey: processEnvironment[\"UMANS_AI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_API_KEY\"]\n)\nlet session = provider.model(\"umans-kimi-k2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-deepseek-v4-flash-0731": {
"id": "umans-deepseek-v4-flash-0731",
"name": "DeepSeek V4 Flash",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393215
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai/umans-deepseek-v4-flash-0731\", apiKey: processEnvironment[\"UMANS_AI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_API_KEY\"]\n)\nlet session = provider.model(\"umans-deepseek-v4-flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-coder": {
"id": "umans-coder",
"name": "Umans Coder",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.19
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai/umans-coder\", apiKey: processEnvironment[\"UMANS_AI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_API_KEY\"]\n)\nlet session = provider.model(\"umans-coder\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-flash": {
"id": "umans-flash",
"name": "Umans Flash",
"description": "Open multimodal Qwen MoE for local agents that need vision, audio, and code",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.15,
"output": 1,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai/umans-flash\", apiKey: processEnvironment[\"UMANS_AI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_API_KEY\"]\n)\nlet session = provider.model(\"umans-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"umans-deepseek-v4-pro-0813": {
"id": "umans-deepseek-v4-pro-0813",
"name": "DeepSeek V4 Pro",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 393215
},
"cost": {
"input": 1.32,
"output": 3.96,
"cache_read": 0.044
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"umans-ai/umans-deepseek-v4-pro-0813\", apiKey: processEnvironment[\"UMANS_AI_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.code.umans.ai/v1\")!,\n apiKey: processEnvironment[\"UMANS_AI_API_KEY\"]\n)\nlet session = provider.model(\"umans-deepseek-v4-pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"sakana": {
"id": "sakana",
"name": "Sakana AI",
"baseURL": "https://api.sakana.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"SAKANA_API_KEY"
],
"doc": "https://console.sakana.ai/models",
"modelCount": 4,
"models": {
"fugu": {
"id": "fugu",
"name": "Fugu",
"description": "Multi-agent model for routing expert agents across complex analytical tasks",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"provider": {
"shape": "responses"
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sakana/fugu\", apiKey: processEnvironment[\"SAKANA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.sakana.ai/v1\")!,\n apiKey: processEnvironment[\"SAKANA_API_KEY\"]\n)\nlet session = provider.model(\"fugu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fugu-ultra": {
"id": "fugu-ultra",
"name": "Fugu Ultra",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"provider": {
"shape": "responses"
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sakana/fugu-ultra\", apiKey: processEnvironment[\"SAKANA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.sakana.ai/v1\")!,\n apiKey: processEnvironment[\"SAKANA_API_KEY\"]\n)\nlet session = provider.model(\"fugu-ultra\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"fugu-ultra-20260615": {
"id": "fugu-ultra-20260615",
"name": "Fugu Ultra",
"description": "Quality-first multi-agent model for hard research, analysis, and competitions",
"family": "fugu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-15",
"last_updated": "2026-06-15",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 1000000
},
"provider": {
"shape": "responses"
},
"cost": {
"input": 5,
"output": 30,
"cache_read": 0.5,
"tiers": [
{
"input": 10,
"output": 45,
"cache_read": 1,
"tier": {
"type": "context",
"size": 272000
}
}
],
"context_over_200k": {
"input": 10,
"output": 45,
"cache_read": 1
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sakana/fugu-ultra-20260615\", apiKey: processEnvironment[\"SAKANA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.sakana.ai/v1\")!,\n apiKey: processEnvironment[\"SAKANA_API_KEY\"]\n)\nlet session = provider.model(\"fugu-ultra-20260615\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"sakana-namazu": {
"id": "sakana-namazu",
"name": "Sakana Namazu",
"description": "Japanese-specialized reasoning model based on Kimi K2.6 and tuned for Japanese language, culture, and business workflows",
"family": "sakana-namazu",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.95,
"output": 4,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"sakana/sakana-namazu\", apiKey: processEnvironment[\"SAKANA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.sakana.ai/v1\")!,\n apiKey: processEnvironment[\"SAKANA_API_KEY\"]\n)\nlet session = provider.model(\"sakana-namazu\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"deepinfra": {
"id": "deepinfra",
"name": "Deep Infra",
"baseURL": "",
"npm": "@ai-sdk/deepinfra",
"swiftDriver": "openaiChat",
"env": [
"DEEPINFRA_API_KEY"
],
"doc": "https://deepinfra.com/models",
"modelCount": 67,
"models": {
"ByteDance/Seed-2.0-mini": {
"id": "ByteDance/Seed-2.0-mini",
"name": "Seed 2.0 Mini",
"description": "Lightweight ByteDance Seed 2.0 model for low-latency multimodal reasoning and high-volume tasks",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 32000
},
"cost": {
"input": 0.1,
"output": 0.4,
"cache_read": 0.02,
"tiers": [
{
"input": 0.2,
"output": 0.8,
"cache_read": 0.2,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/ByteDance/Seed-2.0-mini\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"ByteDance/Seed-2.0-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ByteDance/Seed-2.0-code": {
"id": "ByteDance/Seed-2.0-code",
"name": "Seed 2.0 Code",
"description": "ByteDance Seed coding model for multimodal software engineering and long-running agents",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 131072
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.1,
"tiers": [
{
"input": 1,
"output": 6,
"cache_read": 0.2,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/ByteDance/Seed-2.0-code\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"ByteDance/Seed-2.0-code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"ByteDance/Seed-2.0-pro": {
"id": "ByteDance/Seed-2.0-pro",
"name": "Seed 2.0 Pro",
"description": "Flagship ByteDance Seed 2.0 model for complex multimodal reasoning and long-horizon agent workflows",
"family": "seed",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-14",
"last_updated": "2026-02-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 128000
},
"cost": {
"input": 0.5,
"output": 3,
"cache_read": 0.1,
"tiers": [
{
"input": 1,
"output": 6,
"cache_read": 0.2,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/ByteDance/Seed-2.0-pro\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"ByteDance/Seed-2.0-pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"stepfun-ai/Step-3.7-Flash": {
"id": "stepfun-ai/Step-3.7-Flash",
"name": "Step 3.7 Flash",
"description": "Newer StepFun flash model for faster agents, coding, and multimodal prompts",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"knowledge": "2026-03-01",
"release_date": "2026-05-29",
"last_updated": "2026-05-29",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 256000
},
"cost": {
"input": 0.2,
"output": 1.15,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/stepfun-ai/Step-3.7-Flash\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"stepfun-ai/Step-3.7-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3": {
"id": "deepseek-ai/DeepSeek-V3",
"name": "DeepSeek-V3",
"description": "Open DeepSeek MoE chat model for coding, math, and general reasoning",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-12-26",
"last_updated": "2024-12-26",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 8192
},
"cost": {
"input": 0.32,
"output": 0.89
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/deepseek-ai/DeepSeek-V3\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Flash": {
"id": "deepseek-ai/DeepSeek-V4-Flash",
"name": "DeepSeek V4 Flash",
"description": "Fast DeepSeek V4 lane for economical reasoning, coding, and long-context work",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 16384
},
"cost": {
"input": 0.09,
"output": 0.18,
"cache_read": 0.018
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/deepseek-ai/DeepSeek-V4-Flash\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3-0324": {
"id": "deepseek-ai/DeepSeek-V3-0324",
"name": "DeepSeek V3 0324",
"description": "March 2025 checkpoint of DeepSeek-V3 with improved reasoning and coding",
"family": "deepseek",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-03-24",
"last_updated": "2025-03-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 163840
},
"cost": {
"input": 0.24,
"output": 0.9,
"cache_read": 0.135
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/deepseek-ai/DeepSeek-V3-0324\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3-0324\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Flash-Vision-Exp": {
"id": "deepseek-ai/DeepSeek-V4-Flash-Vision-Exp",
"name": "DeepSeek V4 Flash Vision Exp",
"description": "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-21",
"last_updated": "2026-08-21",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.44,
"output": 1.32,
"cache_read": 0.014
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/deepseek-ai/DeepSeek-V4-Flash-Vision-Exp\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash-Vision-Exp\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Flash-0731": {
"id": "deepseek-ai/DeepSeek-V4-Flash-0731",
"name": "DeepSeek V4 Flash 0731",
"description": "Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding",
"family": "deepseek-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-07-31",
"last_updated": "2026-07-31",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.06,
"output": 0.18,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/deepseek-ai/DeepSeek-V4-Flash-0731\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Flash-0731\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.1": {
"id": "deepseek-ai/DeepSeek-V3.1",
"name": "DeepSeek-V3.1",
"description": "Hybrid-reasoning DeepSeek model with thinking and non-thinking modes",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-21",
"last_updated": "2025-08-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 163840,
"output": 8192
},
"cost": {
"input": 0.25,
"output": 0.95,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/deepseek-ai/DeepSeek-V3.1\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-R1-0528": {
"id": "deepseek-ai/DeepSeek-R1-0528",
"name": "DeepSeek-R1-0528",
"description": "DeepSeek reasoning model for multi-step analysis, math, coding, and tools",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-07",
"release_date": "2025-05-28",
"last_updated": "2025-05-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 163840,
"output": 64000
},
"cost": {
"input": 0.5,
"output": 2.15,
"cache_read": 0.35
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/deepseek-ai/DeepSeek-R1-0528\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-R1-0528\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4.1-Flash": {
"id": "deepseek-ai/DeepSeek-V4.1-Flash",
"name": "DeepSeek V4.1 Flash",
"description": "DeepSeek V4.1 Flash model for reasoning and agentic coding",
"family": "deepseek-flash",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-09-10",
"last_updated": "2026-09-10",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 0.2,
"output": 0.6,
"cache_read": 0.006
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/deepseek-ai/DeepSeek-V4.1-Flash\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4.1-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro-0813": {
"id": "deepseek-ai/DeepSeek-V4-Pro-0813",
"name": "DeepSeek V4 Pro 0813",
"description": "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 384000
},
"cost": {
"input": 1.3,
"output": 2.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/deepseek-ai/DeepSeek-V4-Pro-0813\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro-0813\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V3.2": {
"id": "deepseek-ai/DeepSeek-V3.2",
"name": "DeepSeek-V3.2",
"description": "DeepSeek chat model for instruction following, coding, and analysis",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2025-12-02",
"last_updated": "2025-12-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 163840,
"output": 64000
},
"cost": {
"input": 0.26,
"output": 0.38,
"cache_read": 0.13
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/deepseek-ai/DeepSeek-V3.2\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V3.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"deepseek-ai/DeepSeek-V4-Pro": {
"id": "deepseek-ai/DeepSeek-V4-Pro",
"name": "DeepSeek V4 Pro",
"description": "Open MoE flagship with million-token context for coding and long agent runs",
"family": "deepseek-thinking",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-05",
"release_date": "2026-04-24",
"last_updated": "2026-04-24",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 16384
},
"cost": {
"input": 1.3,
"output": 2.6,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/deepseek-ai/DeepSeek-V4-Pro\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"deepseek-ai/DeepSeek-V4-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning": {
"id": "nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning",
"name": "Nemotron 3 Nano Omni 30B A3B Reasoning",
"description": "Open Nemotron omni model combining reasoning with text, vision, and audio",
"family": "nemotron",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-28",
"last_updated": "2026-04-28",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"status": "deprecated",
"cost": {
"input": 0.2,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Llama-3.3-Nemotron-Super-49B-v1.5": {
"id": "nvidia/Llama-3.3-Nemotron-Super-49B-v1.5",
"name": "Llama 3.3 Nemotron Super 49B v1.5",
"description": "Nemotron model for efficient reasoning, coding, and specialized AI agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-25",
"last_updated": "2025-07-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"status": "deprecated",
"cost": {
"input": 0.4,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/nvidia/Llama-3.3-Nemotron-Super-49B-v1.5\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Llama-3.3-Nemotron-Super-49B-v1.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"nvidia/Nemotron-3-Nano-30B-A3B": {
"id": "nvidia/Nemotron-3-Nano-30B-A3B",
"name": "Nemotron 3 Nano 30B A3B",
"description": "Small Nemotron 3 MoE for efficient coding, math, and long-context agents",
"family": "nemotron",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"temperature": true,
"release_date": "2025-12-15",
"last_updated": "2025-12-15",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.05,
"output": 0.2,
"cache_read": 0.025
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/nvidia/Nemotron-3-Nano-30B-A3B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"nvidia/Nemotron-3-Nano-30B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-4b-it": {
"id": "google/gemma-3-4b-it",
"name": "Gemma 3 4B IT",
"description": "Open multimodal Gemma instruction model for efficient text generation and image understanding",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-03-12",
"last_updated": "2025-03-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/google/gemma-3-4b-it\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3-4b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-31B-it": {
"id": "google/gemma-4-31B-it",
"name": "Gemma 4 31B IT",
"description": "Largest Gemma 4 instruction model for open, self-hosted chat and reasoning",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.13,
"output": 0.38
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/google/gemma-4-31B-it\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-31B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-E4B-it": {
"id": "google/gemma-4-E4B-it",
"name": "Gemma 4 E4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.02,
"output": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/google/gemma-4-E4B-it\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-E4B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-27b-it": {
"id": "google/gemma-3-27b-it",
"name": "Gemma 3 27B IT",
"description": "Largest open Gemma 3 instruction model for multilingual text generation and visual understanding",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-03-12",
"last_updated": "2025-03-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.08,
"output": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/google/gemma-3-27b-it\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3-27b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-4-26B-A4B-it": {
"id": "google/gemma-4-26B-A4B-it",
"name": "Gemma 4 26B A4B IT",
"description": "Open Gemma instruction model for efficient chat and self-hosted deployments",
"family": "gemma",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.07,
"output": 0.34
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/google/gemma-4-26B-A4B-it\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-4-26B-A4B-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"google/gemma-3-12b-it": {
"id": "google/gemma-3-12b-it",
"name": "Gemma 3 12B IT",
"description": "Open multimodal Gemma instruction model for multilingual text generation and image understanding",
"family": "gemma",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-08",
"release_date": "2025-03-12",
"last_updated": "2025-03-12",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 131072
},
"cost": {
"input": 0.05,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/google/gemma-3-12b-it\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"google/gemma-3-12b-it\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.1": {
"id": "zai-org/GLM-5.1",
"name": "GLM-5.1",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-07",
"last_updated": "2026-04-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 16384
},
"cost": {
"input": 1.05,
"output": 3.5,
"cache_read": 0.205
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/zai-org/GLM-5.1\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.3": {
"id": "zai-org/GLM-5.3",
"name": "GLM-5.3",
"description": "Flagship GLM model for long-horizon coding, agents, and complex project delivery",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/zai-org/GLM-5.3\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.2": {
"id": "zai-org/GLM-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 32768
},
"cost": {
"input": 0.75,
"output": 2.4,
"cache_read": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/zai-org/GLM-5.2\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.7-Flash": {
"id": "zai-org/GLM-4.7-Flash",
"name": "GLM-4.7-Flash",
"description": "Efficient GLM model for fast reasoning, coding, and agent workflows",
"family": "glm-flash",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-01-19",
"last_updated": "2026-01-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 16384
},
"status": "deprecated",
"cost": {
"input": 0.06,
"output": 0.4,
"cache_read": 0.01
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/zai-org/GLM-4.7-Flash\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.7-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.7": {
"id": "zai-org/GLM-4.7",
"name": "GLM-4.7",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-12-22",
"last_updated": "2025-12-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 16384
},
"cost": {
"input": 0.4,
"output": 1.75,
"cache_read": 0.08
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/zai-org/GLM-4.7\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5": {
"id": "zai-org/GLM-5",
"name": "GLM-5",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-12",
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 16384
},
"status": "deprecated",
"cost": {
"input": 0.6,
"output": 2.08,
"cache_read": 0.12
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/zai-org/GLM-5\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-4.6": {
"id": "zai-org/GLM-4.6",
"name": "GLM-4.6",
"description": "Flagship GLM model for hybrid reasoning, coding, and agentic engineering",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-30",
"last_updated": "2025-09-30",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 0.5,
"output": 2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/zai-org/GLM-4.6\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-4.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"zai-org/GLM-5.3-Flash": {
"id": "zai-org/GLM-5.3-Flash",
"name": "GLM-5.3-Flash",
"description": "Native multimodal GLM model for efficient coding and long-horizon agent tasks",
"family": "glm",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.5,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/zai-org/GLM-5.3-Flash\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"zai-org/GLM-5.3-Flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/Inkling-Small": {
"id": "thinkingmachines/Inkling-Small",
"name": "Inkling Small",
"description": "Multimodal MoE reasoning model (276B total, 12B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-30",
"last_updated": "2026-07-30",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 1048576
},
"cost": {
"input": 0.45,
"output": 1.2,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/thinkingmachines/Inkling-Small\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/Inkling-Small\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"thinkingmachines/Inkling": {
"id": "thinkingmachines/Inkling",
"name": "Inkling",
"description": "Multimodal MoE reasoning model (975B total, 41B active) for text, image, and audio",
"family": "ling",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 1048576
},
"cost": {
"input": 0.95,
"output": 4.05,
"cache_read": 0.16
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/thinkingmachines/Inkling\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"thinkingmachines/Inkling\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.7-Max": {
"id": "Qwen/Qwen3.7-Max",
"name": "Qwen3.7 Max",
"description": "Qwen frontier model tuned for agent frameworks, coding assistants, and long tasks",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 2.5,
"output": 7.5,
"cache_read": 0.5,
"tiers": [
{
"input": 5,
"output": 15,
"cache_read": 1,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 6.25,
"output": 18.5,
"cache_read": 1.25,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3.7-Max\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.7-Max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.8-27B": {
"id": "Qwen/Qwen3.8-27B",
"name": "Qwen3.8 27B",
"description": "Dense 27B vision-language model for coding, agent tasks, and image and video understanding",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 3,
"cache_read": 0.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3.8-27B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.8-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-27B": {
"id": "Qwen/Qwen3.5-27B",
"name": "Qwen3.5 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.26,
"output": 2.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3.5-27B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Coder-480B-A35B-Instruct-Turbo": {
"id": "Qwen/Qwen3-Coder-480B-A35B-Instruct-Turbo",
"name": "Qwen3 Coder 480B A35B Instruct Turbo",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 66536
},
"cost": {
"input": 0.3,
"output": 1,
"cache_read": 0.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3-Coder-480B-A35B-Instruct-Turbo\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Coder-480B-A35B-Instruct-Turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.8-Max": {
"id": "Qwen/Qwen3.8-Max",
"name": "Qwen3.8 Max",
"description": "2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-03",
"last_updated": "2026-08-03",
"modalities": {
"input": [
"text",
"image",
"video",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 131072
},
"cost": {
"input": 1.65,
"output": 4.951,
"cache_read": 0.206
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3.8-Max\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.8-Max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-9B": {
"id": "Qwen/Qwen3.5-9B",
"name": "Qwen3.5 9B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.1,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3.5-9B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-9B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-30B-A3B": {
"id": "Qwen/Qwen3-30B-A3B",
"name": "Qwen3 30B A3B",
"description": "Sparse MoE Qwen model with 3B active parameters for efficient chat and reasoning",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-28",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 16384
},
"cost": {
"input": 0.12,
"output": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3-30B-A3B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-30B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-122B-A10B": {
"id": "Qwen/Qwen3.5-122B-A10B",
"name": "Qwen3.5 122B-A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.29,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3.5-122B-A10B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-122B-A10B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.8-2.4T-A95B": {
"id": "Qwen/Qwen3.8-2.4T-A95B",
"name": "Qwen3.8 2.4T A95B",
"description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3.8-2.4T-A95B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.8-2.4T-A95B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-235B-A22B-Instruct-2507": {
"id": "Qwen/Qwen3-235B-A22B-Instruct-2507",
"name": "Qwen3 235B-A22B Instruct 2507",
"description": "Updated large open Qwen3 MoE instruct model for multilingual chat, coding, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-21",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.09,
"output": 0.55
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3-235B-A22B-Instruct-2507\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-235B-A22B-Instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-VL-235B-A22B-Instruct": {
"id": "Qwen/Qwen3-VL-235B-A22B-Instruct",
"name": "Qwen3 VL 235B A22B Instruct",
"description": "Qwen vision-language instruct model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 0.88,
"cache_read": 0.11
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3-VL-235B-A22B-Instruct\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-VL-235B-A22B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Next-80B-A3B-Instruct": {
"id": "Qwen/Qwen3-Next-80B-A3B-Instruct",
"name": "Qwen3-Next 80B-A3B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.09,
"output": 1.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3-Next-80B-A3B-Instruct\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Next-80B-A3B-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-397B-A17B": {
"id": "Qwen/Qwen3.5-397B-A17B",
"name": "Qwen 3.5 397B A17B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-01",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 81920
},
"cost": {
"input": 0.45,
"output": 3,
"cache_read": 0.22
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3.5-397B-A17B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-397B-A17B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.5-35B-A3B": {
"id": "Qwen/Qwen3.5-35B-A3B",
"name": "Qwen 3.5 35B A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-02-01",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 81920
},
"cost": {
"input": 0.14,
"output": 1,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3.5-35B-A3B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.5-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-32B": {
"id": "Qwen/Qwen3-32B",
"name": "Qwen3 32B",
"description": "Dense open Qwen model for self-hosted chat, reasoning, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 16384
},
"cost": {
"input": 0.08,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3-32B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-32B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-27B": {
"id": "Qwen/Qwen3.6-27B",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"video",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.32,
"output": 3.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3.6-27B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-27B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3.6-35B-A3B": {
"id": "Qwen/Qwen3.6-35B-A3B",
"name": "Qwen3.6 35B A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-01",
"last_updated": "2026-04-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 81920
},
"cost": {
"input": 0.1,
"output": 0.95
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3.6-35B-A3B\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3.6-35B-A3B\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Qwen/Qwen3-Max": {
"id": "Qwen/Qwen3-Max",
"name": "Qwen3 Max",
"description": "Flagship Qwen3 model for coding agents, complex reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 256000,
"output": 65536
},
"cost": {
"input": 1.2,
"output": 6,
"cache_read": 0.24,
"tiers": [
{
"input": 2.4,
"output": 12,
"cache_read": 0.48,
"tier": {
"type": "context",
"size": 32000
}
},
{
"input": 3,
"output": 15,
"cache_read": 0.6,
"tier": {
"type": "context",
"size": 128000
}
}
]
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/Qwen/Qwen3-Max\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"Qwen/Qwen3-Max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M2.5": {
"id": "MiniMaxAI/MiniMax-M2.5",
"name": "MiniMax M2.5",
"description": "MiniMax model for chat, coding, office work, and agentic tasks",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"temperature": true,
"knowledge": "2025-06",
"release_date": "2026-02-12",
"last_updated": "2026-02-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 131072
},
"status": "deprecated",
"cost": {
"input": 0.15,
"output": 1.15,
"cache_read": 0.03
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/MiniMaxAI/MiniMax-M2.5\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M3": {
"id": "MiniMaxAI/MiniMax-M3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 524288,
"output": 512000
},
"cost": {
"input": 0.28,
"output": 1.1,
"cache_read": 0.056
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/MiniMaxAI/MiniMax-M3\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMaxAI/MiniMax-M2.7": {
"id": "MiniMaxAI/MiniMax-M2.7",
"name": "MiniMax-M2.7",
"description": "Open MiniMax flagship for coding agents, office automation, and complex environments",
"family": "minimax",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"temperature": true,
"release_date": "2026-03-18",
"last_updated": "2026-03-18",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 196608,
"output": 131072
},
"status": "deprecated",
"cost": {
"input": 0.25,
"output": 1,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/MiniMaxAI/MiniMax-M2.7\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"MiniMaxAI/MiniMax-M2.7\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-4-Scout-17B-16E-Instruct": {
"id": "meta-llama/Llama-4-Scout-17B-16E-Instruct",
"name": "Llama 4 Scout 17B",
"description": "Open multimodal Llama model for long-context analysis and efficient agents",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 327680,
"output": 16384
},
"cost": {
"input": 0.1,
"output": 0.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/meta-llama/Llama-4-Scout-17B-16E-Instruct\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-4-Scout-17B-16E-Instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-3.3-70B-Instruct-Turbo": {
"id": "meta-llama/Llama-3.3-70B-Instruct-Turbo",
"name": "Llama 3.3 70B Turbo",
"description": "Compact Llama instruction model for fast chat and local deployment",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"release_date": "2024-12-06",
"last_updated": "2024-12-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.1,
"output": 0.32
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/meta-llama/Llama-3.3-70B-Instruct-Turbo\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-3.3-70B-Instruct-Turbo\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8": {
"id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8",
"name": "Llama 4 Maverick 17B FP8",
"description": "Open multimodal Llama model for strong reasoning and fast responses",
"family": "llama",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"release_date": "2025-04-05",
"last_updated": "2025-04-05",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 16384
},
"cost": {
"input": 0.2,
"output": 0.8
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-20b": {
"id": "openai/gpt-oss-20b",
"name": "GPT OSS 20B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.03,
"output": 0.14
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/openai/gpt-oss-20b\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-20b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"openai/gpt-oss-120b": {
"id": "openai/gpt-oss-120b",
"name": "GPT OSS 120B",
"description": "Open-weight GPT model for self-hosted reasoning and instruction-following workloads",
"family": "gpt-oss",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-08-05",
"last_updated": "2025-08-05",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 16384
},
"cost": {
"input": 0.037,
"output": 0.17
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/openai/gpt-oss-120b\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"openai/gpt-oss-120b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.5": {
"id": "moonshotai/Kimi-K2.5",
"name": "Kimi K2.5",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2025-01",
"release_date": "2026-01-27",
"last_updated": "2026-01-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 32768
},
"status": "deprecated",
"cost": {
"input": 0.45,
"output": 2.25,
"cache_read": 0.07
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/moonshotai/Kimi-K2.5\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.7-Code": {
"id": "moonshotai/Kimi-K2.7-Code",
"name": "Kimi K2.7 Code",
"description": "Coding-focused Kimi model, stronger on long-horizon repo work with less overthinking",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-06-12",
"last_updated": "2026-06-12",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 262144
},
"cost": {
"input": 0.68,
"output": 3.4,
"cache_read": 0.136
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/moonshotai/Kimi-K2.7-Code\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.7-Code\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K2.6": {
"id": "moonshotai/Kimi-K2.6",
"name": "Kimi K2.6",
"description": "Kimi multimodal agent model for visual understanding, coding, and planning",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.75,
"output": 3.5,
"cache_read": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/moonshotai/Kimi-K2.6\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"moonshotai/Kimi-K3": {
"id": "moonshotai/Kimi-K3",
"name": "Kimi K3",
"description": "Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work",
"family": "kimi-k3",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-07-16",
"last_updated": "2026-07-16",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 2.85,
"output": 14.25,
"cache_read": 0.285
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/moonshotai/Kimi-K3\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"moonshotai/Kimi-K3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"tencent/Hy3": {
"id": "tencent/Hy3",
"name": "Hy3",
"description": "Tencent Hy reasoning model for coding, instruction following, and agent tasks",
"family": "Hy",
"attachment": false,
"reasoning": true,
"reasoning_options": [],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-06",
"last_updated": "2026-07-06",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"input": 192000,
"output": 128000
},
"cost": {
"input": 0.14,
"output": 0.58,
"cache_read": 0.035
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/tencent/Hy3\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"tencent/Hy3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"XiaomiMiMo/MiMo-V2.5-Pro": {
"id": "XiaomiMiMo/MiMo-V2.5-Pro",
"name": "MiMo-V2.5-Pro",
"description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"audio"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 16384
},
"cost": {
"input": 1,
"output": 3,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/XiaomiMiMo/MiMo-V2.5-Pro\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"XiaomiMiMo/MiMo-V2.5-Pro\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"XiaomiMiMo/MiMo-V2.5": {
"id": "XiaomiMiMo/MiMo-V2.5",
"name": "MiMo-V2.5",
"description": "Open MiMo model for multimodal coding agents and long-context automation",
"family": "mimo",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"knowledge": "2024-12",
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"audio",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.14,
"output": 0.28,
"cache_read": 0.0028
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"deepinfra/XiaomiMiMo/MiMo-V2.5\", apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"\")!,\n apiKey: processEnvironment[\"DEEPINFRA_API_KEY\"]\n)\nlet session = provider.model(\"XiaomiMiMo/MiMo-V2.5\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"wafer.ai": {
"id": "wafer.ai",
"name": "Wafer",
"baseURL": "https://pass.wafer.ai/v1",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"WAFER_API_KEY"
],
"doc": "https://docs.wafer.ai/wafer-pass",
"modelCount": 5,
"models": {
"GLM-5.1": {
"id": "GLM-5.1",
"name": "GLM-5.1",
"description": "General Language Model 5.1 — high-quality bilingual (EN/ZH) generation with strong coding and reasoning capabilities.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-07",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 202752,
"output": 131072
},
"cost": {
"input": 1,
"output": 3.2,
"cache_read": 0.1,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wafer.ai/GLM-5.1\", apiKey: processEnvironment[\"WAFER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://pass.wafer.ai/v1\")!,\n apiKey: processEnvironment[\"WAFER_API_KEY\"]\n)\nlet session = provider.model(\"GLM-5.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"glm5.2-fast": {
"id": "glm5.2-fast",
"name": "GLM5.2-Fast",
"description": "The same model served for high TPS.",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-13",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 3,
"output": 10.25,
"cache_read": 0.5,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wafer.ai/glm5.2-fast\", apiKey: processEnvironment[\"WAFER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://pass.wafer.ai/v1\")!,\n apiKey: processEnvironment[\"WAFER_API_KEY\"]\n)\nlet session = provider.model(\"glm5.2-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"GLM-5.2": {
"id": "GLM-5.2",
"name": "GLM-5.2",
"description": "Open flagship GLM for long-horizon coding agents and million-token context work",
"family": "glm",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"interleaved": {
"field": "reasoning_content"
},
"structured_output": true,
"temperature": true,
"release_date": "2026-06-13",
"last_updated": "2026-06-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 131072
},
"cost": {
"input": 1.2,
"output": 4.1,
"cache_read": 0.2,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wafer.ai/GLM-5.2\", apiKey: processEnvironment[\"WAFER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://pass.wafer.ai/v1\")!,\n apiKey: processEnvironment[\"WAFER_API_KEY\"]\n)\nlet session = provider.model(\"GLM-5.2\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"Kimi-K2.6": {
"id": "Kimi-K2.6",
"name": "Kimi K2.6",
"description": "Kimi K2.6 sparse MoE model with a 262K context window. Available serverless and not included in standard Wafer Pass. Non-ZDR only: requests with `Wafer-ZDR: required` are rejected.",
"family": "kimi-k2",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-01",
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 1.14,
"output": 4.8,
"cache_read": 0.19,
"cache_write": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wafer.ai/Kimi-K2.6\", apiKey: processEnvironment[\"WAFER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://pass.wafer.ai/v1\")!,\n apiKey: processEnvironment[\"WAFER_API_KEY\"]\n)\nlet session = provider.model(\"Kimi-K2.6\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"MiniMax-M3": {
"id": "MiniMax-M3",
"name": "MiniMax-M3",
"description": "MiniMax multimodal model for long-context coding, perception, and agent planning",
"family": "minimax",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "toggle"
},
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-06-01",
"last_updated": "2026-06-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1048576,
"output": 512000
},
"cost": {
"input": 0.33,
"output": 1.32,
"cache_read": 0.07,
"cache_write": 0,
"tiers": [
{
"input": 0.66,
"output": 2.64,
"cache_read": 0.13,
"cache_write": 0,
"tier": {
"type": "context",
"size": 512000
}
}
],
"context_over_200k": {
"input": 0.66,
"output": 2.64,
"cache_read": 0.13,
"cache_write": 0
}
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"wafer.ai/MiniMax-M3\", apiKey: processEnvironment[\"WAFER_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://pass.wafer.ai/v1\")!,\n apiKey: processEnvironment[\"WAFER_API_KEY\"]\n)\nlet session = provider.model(\"MiniMax-M3\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
}
}
},
"kilo": {
"id": "kilo",
"name": "Kilo Gateway",
"baseURL": "https://api.kilo.ai/api/gateway",
"npm": "@ai-sdk/openai-compatible",
"swiftDriver": "openaiChat",
"env": [
"KILO_API_KEY"
],
"doc": "https://kilo.ai",
"modelCount": 375,
"models": {
"qwen/qwen3.7-max": {
"id": "qwen/qwen3.7-max",
"name": "Qwen3.7 Max",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-05-21",
"last_updated": "2026-05-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 1.25,
"output": 3.75,
"cache_read": 0.125,
"cache_write": 1.5625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.7-max\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-plus": {
"id": "qwen/qwen3-coder-plus",
"name": "Qwen3 Coder Plus",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.65,
"output": 3.25,
"cache_read": 0.13,
"cache_write": 0.8125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-coder-plus\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-next-80b-a3b-thinking": {
"id": "qwen/qwen3-next-80b-a3b-thinking",
"name": "Qwen3-Next 80B-A3B (Thinking)",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 235929
},
"cost": {
"input": 0.15,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-next-80b-a3b-thinking\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-next-80b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b-thinking-2507": {
"id": "qwen/qwen3-235b-a22b-thinking-2507",
"name": "Qwen: Qwen3 235B A22B Thinking 2507",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-25",
"last_updated": "2025-07-25",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 117964
},
"cost": {
"input": 0.23,
"output": 2.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-235b-a22b-thinking-2507\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b-thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-9b": {
"id": "qwen/qwen3.5-9b",
"name": "Qwen3.5 9B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 235929
},
"cost": {
"input": 0.1,
"output": 0.15
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.5-9b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-9b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-next-80b-a3b-instruct": {
"id": "qwen/qwen3-next-80b-a3b-instruct",
"name": "Qwen3-Next 80B-A3B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09",
"last_updated": "2025-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.0975,
"output": 0.78
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-next-80b-a3b-instruct\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-next-80b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-flash": {
"id": "qwen/qwen3-coder-flash",
"name": "Qwen3 Coder Flash",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-07-28",
"last_updated": "2025-07-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.195,
"output": 0.975,
"cache_read": 0.039,
"cache_write": 0.24375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-coder-flash\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-14b": {
"id": "qwen/qwen3-14b",
"name": "Qwen: Qwen3 14B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-28",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.2275,
"output": 0.91
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-14b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-14b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-plus": {
"id": "qwen/qwen3.6-plus",
"name": "Qwen3.6 Plus",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-02",
"last_updated": "2026-04-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.325,
"output": 1.95,
"cache_write": 0.40625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.6-plus\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-27b": {
"id": "qwen/qwen3.5-27b",
"name": "Qwen3.5 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.195,
"output": 1.56
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.5-27b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-27b": {
"id": "qwen/qwen3.8-27b",
"name": "Qwen3.8 27B",
"description": "Qwen3.8 27B is an open-weight dense vision-language model from Qwen. It is suited for coding, professional workflows, research, multimodal interaction, and long-running agent tasks, with flexible thinking that can be...",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.425,
"output": 2.55,
"cache_read": 0.085,
"cache_write": 0.53125
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.8-27b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-35b-a3b": {
"id": "qwen/qwen3.5-35b-a3b",
"name": "Qwen3.5 35B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 256000,
"output": 16384
},
"cost": {
"input": 0.1625,
"output": 1.3
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.5-35b-a3b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-plus-20260420": {
"id": "qwen/qwen3.5-plus-20260420",
"name": "Qwen: Qwen3.5 Plus 2026-04-20",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.3,
"output": 1.8,
"cache_write": 0.375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.5-plus-20260420\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-plus-20260420\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-32b": {
"id": "qwen/qwen3-32b",
"name": "Qwen3 32B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 16384
},
"cost": {
"input": 0.08,
"output": 0.28
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-32b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-32b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-plus-02-15": {
"id": "qwen/qwen3.5-plus-02-15",
"name": "Qwen: Qwen3.5 Plus 2026-02-15",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-16",
"last_updated": "2026-02-16",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.26,
"output": 1.56
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.5-plus-02-15\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-plus-02-15\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-plus-2025-07-28": {
"id": "qwen/qwen-plus-2025-07-28",
"name": "Qwen: Qwen Plus 0728",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-09-08",
"last_updated": "2025-09-08",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.26,
"output": 0.78
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen-plus-2025-07-28\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-plus-2025-07-28\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder": {
"id": "qwen/qwen3-coder",
"name": "Qwen: Qwen3 Coder 480B A35B",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-23",
"last_updated": "2025-07-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.975,
"output": 4.875
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-coder\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen2.5-vl-72b-instruct": {
"id": "qwen/qwen2.5-vl-72b-instruct",
"name": "Qwen: Qwen2.5 VL 72B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2025-02-01",
"last_updated": "2025-02-01",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 128000,
"output": 115200
},
"cost": {
"input": 0.8,
"output": 1,
"cache_read": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen2.5-vl-72b-instruct\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen2.5-vl-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-next": {
"id": "qwen/qwen3-coder-next",
"name": "Qwen3 Coder Next",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-09",
"release_date": "2026-02-03",
"last_updated": "2026-02-03",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 235929
},
"cost": {
"input": 0.3,
"output": 1.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-coder-next\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-next\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-coder-30b-a3b-instruct": {
"id": "qwen/qwen3-coder-30b-a3b-instruct",
"name": "Qwen3-Coder 30B-A3B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 235929
},
"cost": {
"input": 0.2925,
"output": 1.4625
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-coder-30b-a3b-instruct\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-coder-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b-2507": {
"id": "qwen/qwen3-235b-a22b-2507",
"name": "Qwen: Qwen3 235B A22B Instruct 2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-21",
"last_updated": "2025-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 235929
},
"cost": {
"input": 0.1495,
"output": 0.598
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-235b-a22b-2507\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-flash-02-23": {
"id": "qwen/qwen3.5-flash-02-23",
"name": "Qwen: Qwen3.5-Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.5",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-25",
"last_updated": "2026-02-25",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.065,
"output": 0.26
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.5-flash-02-23\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-flash-02-23\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-397b-a17b": {
"id": "qwen/qwen3.5-397b-a17b",
"name": "Qwen3.5 397B-A17B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-15",
"last_updated": "2026-02-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 235929
},
"cost": {
"input": 0.39,
"output": 2.34
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.5-397b-a17b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-397b-a17b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-8b-thinking": {
"id": "qwen/qwen3-vl-8b-thinking",
"name": "Qwen: Qwen3 VL 8B Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-14",
"last_updated": "2025-10-14",
"modalities": {
"input": [
"image",
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.18,
"output": 2.1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-vl-8b-thinking\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-8b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-27b": {
"id": "qwen/qwen3.6-27b",
"name": "Qwen3.6 27B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-22",
"last_updated": "2026-04-22",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.45,
"output": 2.7
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.6-27b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-27b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.7-flash": {
"id": "qwen/qwen3.7-flash",
"name": "Qwen3.7 Flash",
"description": "Qwen3.7 Flash is a vision-language reasoning model from Alibaba. It is suited for multimodal agents, visual coding, search, and computer interaction, with strengths in object recognition, spatial understanding, and real-world...",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-15",
"last_updated": "2026-07-15",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"input": 991000,
"output": 65536
},
"cost": {
"input": 0.03,
"output": 0.13,
"cache_read": 0.006,
"cache_write": 0.038
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.7-flash\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-30b-a3b-thinking-2507": {
"id": "qwen/qwen3-30b-a3b-thinking-2507",
"name": "Qwen: Qwen3 30B A3B Thinking 2507",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-08-28",
"last_updated": "2025-08-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 81920,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-30b-a3b-thinking-2507\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-30b-a3b-thinking-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-35b-a3b": {
"id": "qwen/qwen3.6-35b-a3b",
"name": "Qwen3.6 35B-A3B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-17",
"last_updated": "2026-04-17",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 235929
},
"cost": {
"input": 0.1,
"output": 0.9,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.6-35b-a3b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-35b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-max-thinking": {
"id": "qwen/qwen3-max-thinking",
"name": "Qwen: Qwen3 Max Thinking",
"description": "Qwen reasoning model for deliberate problem solving, math, and coding",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-09",
"last_updated": "2026-02-09",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.78,
"output": 3.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-max-thinking\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-max-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-max-0902": {
"id": "qwen/qwen3.8-max-0902",
"name": "Qwen3.8 Max 0902",
"description": "Qwen3.8 Max 0902 is an updated snapshot of Qwen3.8 Max from Alibaba's Qwen team. It is a 2.4-trillion-parameter mixture-of-experts model that accepts text, image, and video input and returns text,...",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"minimal",
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-09-02",
"last_updated": "2026-09-02",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.8-max-0902\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-max-0902\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-max": {
"id": "qwen/qwen3-max",
"name": "Qwen3 Max",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.78,
"output": 3.9,
"cache_read": 0.156,
"cache_write": 0.975
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-max\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-max\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-8b-instruct": {
"id": "qwen/qwen3-vl-8b-instruct",
"name": "Qwen: Qwen3 VL 8B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-14",
"last_updated": "2025-10-14",
"modalities": {
"input": [
"image",
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.117,
"output": 0.455
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-vl-8b-instruct\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-8b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-2.4t-a95b": {
"id": "qwen/qwen3.8-2.4t-a95b",
"name": "Qwen3.8 2.4T A95B",
"description": "Open-weight sparse MoE (2.4T total, 95B active), the open-weight twin of Qwen3.8 Max for coding, research, complex reasoning, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-12",
"last_updated": "2026-08-12",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.25,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.8-2.4t-a95b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-2.4t-a95b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-30b-a3b-instruct": {
"id": "qwen/qwen3-vl-30b-a3b-instruct",
"name": "Qwen: Qwen3 VL 30B A3B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 16384
},
"cost": {
"input": 0.13,
"output": 0.52
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-vl-30b-a3b-instruct\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-30b-a3b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-plus": {
"id": "qwen/qwen-plus",
"name": "Qwen Plus",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2024-04",
"release_date": "2024-01-25",
"last_updated": "2025-09-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 32768
},
"cost": {
"input": 0.26,
"output": 0.78,
"cache_read": 0.052,
"cache_write": 0.325
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen-plus\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.5-122b-a10b": {
"id": "qwen/qwen3.5-122b-a10b",
"name": "Qwen3.5 122B-A10B",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 0.26,
"output": 2.08
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.5-122b-a10b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.5-122b-a10b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-2.5-coder-32b-instruct": {
"id": "qwen/qwen-2.5-coder-32b-instruct",
"name": "Qwen2.5 Coder 32B Instruct",
"description": "Qwen coding model for software agents, repository edits, and code reasoning",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2024-11-11",
"last_updated": "2024-11-11",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 29491
},
"cost": {
"input": 0.66,
"output": 1
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen-2.5-coder-32b-instruct\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-2.5-coder-32b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-30b-a3b-instruct-2507": {
"id": "qwen/qwen3-30b-a3b-instruct-2507",
"name": "Qwen: Qwen3 30B A3B Instruct 2507",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-29",
"last_updated": "2025-07-29",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 235929
},
"cost": {
"input": 0.13,
"output": 0.52
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-30b-a3b-instruct-2507\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-30b-a3b-instruct-2507\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-flash": {
"id": "qwen/qwen3.6-flash",
"name": "Qwen3.6 Flash",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen3.6",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 65536
},
"cost": {
"input": 0.1875,
"output": 1.125,
"cache_write": 0.234375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.6-flash\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-30b-a3b": {
"id": "qwen/qwen3-30b-a3b",
"name": "Qwen3 30B A3B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-04-28",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 40960,
"output": 16384
},
"cost": {
"input": 0.13,
"output": 0.52
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-30b-a3b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-30b-a3b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-2.5-7b-instruct": {
"id": "qwen/qwen-2.5-7b-instruct",
"name": "Qwen: Qwen2.5 7B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-10-16",
"last_updated": "2024-10-16",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 29491
},
"cost": {
"input": 0.1,
"output": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen-2.5-7b-instruct\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-2.5-7b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.8-flash": {
"id": "qwen/qwen3.8-flash",
"name": "Qwen3.8 Flash",
"description": "Qwen3.8 Flash is a multimodal reasoning model from Alibaba. It is suited for coding assistance, agentic workflows, visual understanding, document and codebase analysis, desktop interaction, chart analysis, and long-video analysis.",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-26",
"last_updated": "2026-08-26",
"modalities": {
"input": [
"text",
"image",
"video"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.15,
"output": 0.47,
"cache_read": 0.016,
"cache_write": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.8-flash\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.8-flash\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-30b-a3b-thinking": {
"id": "qwen/qwen3-vl-30b-a3b-thinking",
"name": "Qwen: Qwen3 VL 30B A3B Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-06",
"last_updated": "2025-10-06",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.2,
"output": 2.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-vl-30b-a3b-thinking\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-30b-a3b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.6-max-preview": {
"id": "qwen/qwen3.6-max-preview",
"name": "Qwen3.6 Max Preview",
"description": "Flagship Qwen model for complex reasoning, coding, and agentic workflows",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-04-20",
"last_updated": "2026-04-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 65536
},
"cost": {
"input": 1.027,
"output": 6.162,
"cache_write": 1.28375
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.6-max-preview\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.6-max-preview\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen-2.5-72b-instruct": {
"id": "qwen/qwen-2.5-72b-instruct",
"name": "Qwen2.5 72B Instruct",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2024-09-19",
"last_updated": "2024-09-19",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 16384
},
"cost": {
"input": 0.36,
"output": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen-2.5-72b-instruct\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen-2.5-72b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-8b": {
"id": "qwen/qwen3-8b",
"name": "Qwen: Qwen3 8B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2025-04-28",
"last_updated": "2025-04-28",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.117,
"output": 0.455
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-8b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-235b-a22b": {
"id": "qwen/qwen3-235b-a22b",
"name": "Qwen3 235B-A22B",
"description": "Qwen instruction model for multilingual chat, reasoning, and tool use",
"family": "qwen",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2025-04",
"last_updated": "2025-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 8192
},
"cost": {
"input": 0.455,
"output": 1.82
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-235b-a22b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-235b-a22b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-235b-a22b-thinking": {
"id": "qwen/qwen3-vl-235b-a22b-thinking",
"name": "Qwen3 VL 235B A22B Thinking",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.4,
"output": 4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-vl-235b-a22b-thinking\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-235b-a22b-thinking\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-235b-a22b-instruct": {
"id": "qwen/qwen3-vl-235b-a22b-instruct",
"name": "Qwen3 VL 235B A22B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-03-31",
"release_date": "2025-09-23",
"last_updated": "2025-09-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": true,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.26,
"output": 1.04
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-vl-235b-a22b-instruct\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-235b-a22b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3.7-plus": {
"id": "qwen/qwen3.7-plus",
"name": "Qwen3.7 Plus",
"description": "Qwen3.7-Plus is a cost-effective model in Alibaba's Qwen3.7 series. It supports text and image input with text output, building on the series' text capabilities with a comprehensive upgrade to its...",
"family": "qwen",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"knowledge": "2025-04",
"release_date": "2026-06-02",
"last_updated": "2026-06-02",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 131072
},
"cost": {
"input": 0.32,
"output": 1.28,
"cache_read": 0.032,
"cache_write": 0.4
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3.7-plus\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3.7-plus\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"qwen/qwen3-vl-32b-instruct": {
"id": "qwen/qwen3-vl-32b-instruct",
"name": "Qwen: Qwen3 VL 32B Instruct",
"description": "Qwen vision-language model for visual reasoning, documents, and agent tasks",
"family": "qwen",
"attachment": true,
"reasoning": false,
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2025-10-23",
"last_updated": "2025-10-23",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.104,
"output": 0.416
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/qwen/qwen3-vl-32b-instruct\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"qwen/qwen3-vl-32b-instruct\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"baidu/ernie-4.5-vl-424b-a47b": {
"id": "baidu/ernie-4.5-vl-424b-a47b",
"name": "Baidu: ERNIE 4.5 VL 424B A47B ",
"description": "Multimodal reasoning model for visual analysis, planning, and tool use",
"family": "ernie",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-06-30",
"last_updated": "2025-06-30",
"modalities": {
"input": [
"image",
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 123000,
"output": 16000
},
"cost": {
"input": 0.42,
"output": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/baidu/ernie-4.5-vl-424b-a47b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"baidu/ernie-4.5-vl-424b-a47b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aion-labs/aion-2.0": {
"id": "aion-labs/aion-2.0",
"name": "AionLabs: Aion-2.0",
"description": "Reasoning model for deliberate analysis, multi-step problem solving, and tool use",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-02-23",
"last_updated": "2026-02-23",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.8,
"output": 1.6,
"cache_read": 0.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/aion-labs/aion-2.0\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"aion-labs/aion-2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aion-labs/aion-rp-llama-3.1-8b": {
"id": "aion-labs/aion-rp-llama-3.1-8b",
"name": "AionLabs: Aion-RP 1.0 (8B)",
"description": "Open Llama instruction model for multilingual chat, reasoning, and coding",
"family": "llama",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-02-04",
"last_updated": "2025-02-04",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 32768,
"output": 29491
},
"cost": {
"input": 0.8,
"output": 1.6
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/aion-labs/aion-rp-llama-3.1-8b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"aion-labs/aion-rp-llama-3.1-8b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aion-labs/aion-3.0": {
"id": "aion-labs/aion-3.0",
"name": "AionLabs: Aion-3.0",
"description": "Aion-3.0 is a multi-model roleplaying and storytelling system from AionLabs, built on the GLM family of models. It uses a collaborative generation process in which multiple specialized models each contribute...",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-07",
"last_updated": "2026-07-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 3,
"output": 6,
"cache_read": 0.75
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/aion-labs/aion-3.0\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"aion-labs/aion-3.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"aion-labs/aion-3.0-mini": {
"id": "aion-labs/aion-3.0-mini",
"name": "AionLabs: Aion-3.0-Mini",
"description": "Aion-3.0 Mini is a multi-model roleplaying and storytelling system from AionLabs, built on the DeepSeek family of models. It uses a collaborative generation process in which multiple specialized models each...",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-07",
"last_updated": "2026-07-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 131072,
"output": 32768
},
"cost": {
"input": 0.7,
"output": 1.4,
"cache_read": 0.18
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/aion-labs/aion-3.0-mini\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"aion-labs/aion-3.0-mini\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"~anthropic/claude-fable-latest": {
"id": "~anthropic/claude-fable-latest",
"name": "Anthropic: Claude Fable Latest ($$$$)",
"description": "This model always redirects to the latest model in the Claude Fable family.",
"family": "claude-fable",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"release_date": "2026-06-09",
"last_updated": "2026-06-09",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 10,
"output": 50,
"cache_read": 0.25,
"cache_write": 12.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/~anthropic/claude-fable-latest\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"~anthropic/claude-fable-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"~anthropic/claude-opus-latest": {
"id": "~anthropic/claude-opus-latest",
"name": "Anthropic: Claude Opus Latest",
"description": "Flagship Claude model for deep reasoning, coding, and long-horizon agents",
"family": "claude-opus",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-21",
"last_updated": "2026-04-21",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 5,
"output": 25,
"cache_read": 0.5,
"cache_write": 6.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/~anthropic/claude-opus-latest\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"~anthropic/claude-opus-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"~anthropic/claude-haiku-latest": {
"id": "~anthropic/claude-haiku-latest",
"name": "Anthropic Claude Haiku Latest",
"description": "Fast Claude model for responsive assistance, classification, and lightweight agents",
"family": "claude-haiku",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 200000,
"output": 64000
},
"cost": {
"input": 1,
"output": 5,
"cache_read": 0.1,
"cache_write": 1.25
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/~anthropic/claude-haiku-latest\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"~anthropic/claude-haiku-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"~anthropic/claude-sonnet-latest": {
"id": "~anthropic/claude-sonnet-latest",
"name": "Anthropic Claude Sonnet Latest",
"description": "Balanced Claude model for coding, analysis, agent workflows, and cost control",
"family": "claude-sonnet",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": false,
"knowledge": "2025-08-31",
"release_date": "2026-04-27",
"last_updated": "2026-04-27",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1000000,
"output": 128000
},
"cost": {
"input": 2,
"output": 10,
"cache_read": 0.2,
"cache_write": 2.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/~anthropic/claude-sonnet-latest\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"~anthropic/claude-sonnet-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"morph/morph-v3-large": {
"id": "morph/morph-v3-large",
"name": "Morph: Morph V3 Large",
"description": "Flagship model for demanding analysis, coding, and production agent workflows",
"family": "morph",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2025-07-07",
"last_updated": "2025-07-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 131072
},
"cost": {
"input": 0.9,
"output": 1.9
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/morph/morph-v3-large\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"morph/morph-v3-large\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"morph/morph-v3-fast": {
"id": "morph/morph-v3-fast",
"name": "Morph: Morph V3 Fast",
"description": "Efficient model for low-latency assistance, extraction, and routine automation",
"family": "morph",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": false,
"temperature": true,
"release_date": "2025-07-07",
"last_updated": "2025-07-07",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 81920,
"output": 38000
},
"cost": {
"input": 0.8,
"output": 1.2
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/morph/morph-v3-fast\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"morph/morph-v3-fast\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"undi95/remm-slerp-l2-13b": {
"id": "undi95/remm-slerp-l2-13b",
"name": "ReMM SLERP 13B",
"description": "Open-weight instruction model for adaptable chat and self-hosted production workloads",
"attachment": false,
"reasoning": false,
"tool_call": false,
"structured_output": true,
"temperature": true,
"release_date": "2023-07-22",
"last_updated": "2023-07-22",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 6144,
"output": 5529
},
"cost": {
"input": 0.35,
"output": 0.65
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/undi95/remm-slerp-l2-13b\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"undi95/remm-slerp-l2-13b\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"~deepseek/deepseek-v4-flash-latest": {
"id": "~deepseek/deepseek-v4-flash-latest",
"name": "DeepSeek V4 Flash Latest",
"description": "This model always redirects to the latest model in the DeepSeek V4 Flash family.",
"family": "deepseek",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"low",
"high",
"max"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-01",
"last_updated": "2026-08-01",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048576,
"output": 943718
},
"cost": {
"input": 0.04,
"output": 0.08,
"cache_read": 0.008
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/~deepseek/deepseek-v4-flash-latest\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"~deepseek/deepseek-v4-flash-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"dots-studio/dots-3-note-preview:free": {
"id": "dots-studio/dots-3-note-preview:free",
"name": "Dots Studio: Dots3-Note Preview (free)",
"description": "Dots3-Note Preview is an open-weight mixture-of-experts model from Dots Studio, with 16B active parameters out of 280B total. It is the lightest model in the Dots 3 family and is...",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-08-14",
"last_updated": "2026-08-14",
"modalities": {
"input": [
"text",
"image"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 512000,
"output": 460800
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/dots-studio/dots-3-note-preview:free\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"dots-studio/dots-3-note-preview:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"~x-ai/grok-latest": {
"id": "~x-ai/grok-latest",
"name": "xAI: Grok Latest",
"description": "This model always redirects to the latest Grok model from xAI.",
"family": "grok",
"attachment": true,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"low",
"medium",
"high",
"xhigh"
]
}
],
"tool_call": true,
"structured_output": true,
"temperature": true,
"release_date": "2026-07-08",
"last_updated": "2026-07-08",
"modalities": {
"input": [
"text",
"image",
"pdf"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 500000,
"output": 450000
},
"cost": {
"input": 2,
"output": 6,
"cache_read": 0.5
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/~x-ai/grok-latest\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"~x-ai/grok-latest\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"meituan/longcat-2.0": {
"id": "meituan/longcat-2.0",
"name": "Meituan: LongCat 2.0",
"description": "LongCat 2.0 is a sparse mixture-of-experts language model from Meituan, with 48B active parameters out of 1.6T total. It is suited for coding, repository-level changes, long-horizon problem solving, and agentic...",
"family": "longcat",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-20",
"last_updated": "2026-07-20",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 1048756,
"output": 262144
},
"cost": {
"input": 0.75,
"output": 3,
"cache_read": 0.015
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/meituan/longcat-2.0\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"meituan/longcat-2.0\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-xs-2.1": {
"id": "poolside/laguna-xs-2.1",
"name": "Poolside: Laguna XS 2.1",
"description": "Laguna XS 2.1 is the latest coding agent model in the 33B-A3B category from [Poolside](https://poolside.ai/) and a step forward from their Laguna XS.2 model (released in April 2026). It combines...",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-02",
"last_updated": "2026-07-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0.1,
"output": 0.2,
"cache_read": 0.05
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/poolside/laguna-xs-2.1\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-xs-2.1\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-xs-2.1:free": {
"id": "poolside/laguna-xs-2.1:free",
"name": "Poolside: Laguna XS 2.1 (free)",
"description": "Laguna XS 2.1 is the latest coding agent model in the 33B-A3B category from [Poolside](https://poolside.ai/) and a step forward from their Laguna XS.2 model (released in April 2026). It combines...",
"family": "laguna",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-02",
"last_updated": "2026-07-02",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/poolside/laguna-xs-2.1:free\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-xs-2.1:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-s-2.1:free": {
"id": "poolside/laguna-s-2.1:free",
"name": "Poolside: Laguna S 2.1 (free)",
"description": "Laguna S 2.1 is the latest coding agent model from [Poolside](). Laguna S 2.1 is a 118B total parameter model with 8B active parameters, scoring 70.2% on Terminal-Bench 2.1 and...",
"family": "laguna-s",
"attachment": false,
"reasoning": true,
"reasoning_options": [
{
"type": "effort",
"values": [
"none",
"high"
]
}
],
"tool_call": true,
"structured_output": false,
"temperature": true,
"release_date": "2026-07-21",
"last_updated": "2026-07-21",
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"open_weights": false,
"limit": {
"context": 262144,
"output": 32768
},
"cost": {
"input": 0,
"output": 0
},
"swiftDriver": "openaiChat",
"reasoningField": null,
"swiftSnippet": "import LingXiAgent\n\n// 1. One-line inference via LingXiAgent Registry\nlet model = LingXiAgent.model(\"kilo/poolside/laguna-s-2.1:free\", apiKey: processEnvironment[\"KILO_API_KEY\"])\n\n// 2. Or initialize via OpenAI-Compatible factory\nlet provider = LingXiProvider.openAICompatible(\n baseURL: URL(string: \"https://api.kilo.ai/api/gateway\")!,\n apiKey: processEnvironment[\"KILO_API_KEY\"]\n)\nlet session = provider.model(\"poolside/laguna-s-2.1:free\")\n\n// 3. Stream responses with reasoning support\nfor try await chunk in session.stream(\"Hello, LingXi!\") {\n if let reasoning = chunk.reasoning {\n print(\"[Thinking] \\(reasoning)\", terminator: \"\")\n }\n print(chunk.text, terminator: \"\")\n}"
},
"poolside/laguna-s-2.1": {
"id": "poolside/laguna-s-2.1",
"name": "Poolside: Laguna S 2.1",
"description": "Laguna S 2.1 is the latest coding agent model from [Poolside](