{
  "version": "0.99.0",
  "embedded_extra": 22,
  "providers": [
    {
      "id": "ollama",
      "name": "Ollama",
      "kind": "local",
      "description": "Local Ollama server (default port 11434) — serves any pulled model.",
      "requires_key": false,
      "api_dialect": "openai-chat",
      "default_model": "qwen3.5:4b",
      "cheap_model": "llama3.2:3b",
      "aliases": [],
      "tags": [
        "local",
        "open-source"
      ],
      "models": [
        {
          "id": "qwen",
          "model": "qwen3.5:4b",
          "context_window_tokens": 128000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": true
          }
        },
        {
          "id": "llama",
          "model": "llama3.2:3b",
          "context_window_tokens": 128000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": true
          }
        }
      ]
    },
    {
      "id": "lmstudio",
      "name": "LM Studio",
      "kind": "local",
      "description": "LM Studio local server (default port 1234) — serves the loaded model.",
      "requires_key": false,
      "api_dialect": "openai-chat",
      "default_model": "qwen3.5-4b",
      "cheap_model": "qwen3.5-4b",
      "aliases": [
        "lm-studio"
      ],
      "tags": [
        "local",
        "open-source"
      ],
      "models": [
        {
          "id": "qwen",
          "model": "qwen3.5-4b",
          "context_window_tokens": 128000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": true
          }
        }
      ]
    },
    {
      "id": "llamacpp",
      "name": "llama.cpp",
      "kind": "local",
      "description": "llama.cpp server (default port 8080) — serves the loaded GGUF; the wire model id is free-form.",
      "requires_key": false,
      "api_dialect": "openai-chat",
      "default_model": "default",
      "cheap_model": "default",
      "aliases": [
        "llama-cpp"
      ],
      "tags": [
        "local",
        "open-source"
      ],
      "models": [
        {
          "id": "default",
          "model": "default",
          "context_window_tokens": 128000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": true
          }
        }
      ]
    },
    {
      "id": "localai",
      "name": "LocalAI",
      "kind": "local",
      "description": "LocalAI server (default port 8080) — an OpenAI-compatible gateway over local backends.",
      "requires_key": false,
      "api_dialect": "openai-chat",
      "default_model": "default",
      "cheap_model": "default",
      "aliases": [],
      "tags": [
        "local",
        "open-source"
      ],
      "models": [
        {
          "id": "default",
          "model": "default",
          "context_window_tokens": 128000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": true
          }
        }
      ]
    },
    {
      "id": "vllm",
      "name": "vLLM",
      "kind": "local",
      "description": "vLLM server (default port 8000) — serves Hugging Face model ids at throughput.",
      "requires_key": false,
      "api_dialect": "openai-chat",
      "default_model": "Qwen/Qwen3-8B",
      "cheap_model": "Qwen/Qwen3-8B",
      "aliases": [],
      "tags": [
        "local",
        "open-source"
      ],
      "models": [
        {
          "id": "qwen",
          "model": "Qwen/Qwen3-8B",
          "context_window_tokens": 128000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": true
          }
        }
      ]
    },
    {
      "id": "mistral",
      "name": "Mistral AI",
      "kind": "cloud",
      "description": "Mistral Large, Medium, Small models.",
      "env_var": "MISTRAL_API_KEY",
      "requires_key": true,
      "api_dialect": "openai-chat",
      "default_model": "mistral-large-latest",
      "cheap_model": "mistral-small-latest",
      "aliases": [],
      "tags": [
        "code",
        "european",
        "frontier",
        "function-calling",
        "open-source",
        "streaming",
        "structured-output"
      ],
      "models": [
        {
          "id": "large",
          "model": "mistral-large-latest",
          "context_window_tokens": 128000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": true,
            "json_mode": "schema"
          }
        },
        {
          "id": "small",
          "model": "mistral-small-latest",
          "context_window_tokens": 32000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": true,
            "json_mode": "schema"
          }
        }
      ]
    },
    {
      "id": "anthropic",
      "name": "Anthropic Claude",
      "kind": "cloud",
      "description": "Claude models (Opus, Sonnet, Haiku).",
      "env_var": "ANTHROPIC_API_KEY",
      "requires_key": true,
      "api_dialect": "anthropic",
      "default_model": "claude-sonnet-4-20250514",
      "cheap_model": "claude-haiku-4-5-20251001",
      "aliases": [
        "claude"
      ],
      "tags": [
        "extended-thinking",
        "frontier",
        "function-calling",
        "long-context",
        "prompt-caching",
        "streaming",
        "structured-output",
        "vision"
      ],
      "models": [
        {
          "id": "sonnet",
          "model": "claude-sonnet-4-20250514",
          "context_window_tokens": 200000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": true,
            "vision": true,
            "json_mode": "schema"
          }
        },
        {
          "id": "haiku",
          "model": "claude-haiku-4-5-20251001",
          "context_window_tokens": 200000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": true,
            "vision": true,
            "json_mode": "schema"
          }
        }
      ]
    },
    {
      "id": "openai",
      "name": "OpenAI",
      "kind": "cloud",
      "description": "GPT-4, GPT-4o, and other OpenAI models.",
      "env_var": "OPENAI_API_KEY",
      "requires_key": true,
      "api_dialect": "openai-chat",
      "default_model": "gpt-4o",
      "cheap_model": "gpt-4o-mini",
      "aliases": [
        "gpt"
      ],
      "tags": [
        "code-execution",
        "frontier",
        "function-calling",
        "image-gen",
        "prompt-caching",
        "realtime",
        "streaming",
        "vision"
      ],
      "models": [
        {
          "id": "gpt-4o",
          "model": "gpt-4o",
          "context_window_tokens": 128000,
          "max_output_tokens": 16384,
          "capabilities": {
            "reasoning": false,
            "vision": true,
            "json_mode": "schema"
          }
        },
        {
          "id": "gpt-4o-mini",
          "model": "gpt-4o-mini",
          "context_window_tokens": 128000,
          "max_output_tokens": 16384,
          "capabilities": {
            "reasoning": false,
            "vision": true,
            "json_mode": "schema"
          }
        }
      ]
    },
    {
      "id": "gemini",
      "name": "Google Gemini",
      "kind": "cloud",
      "description": "Gemini Pro, Flash, and Ultra models.",
      "env_var": "GEMINI_API_KEY",
      "requires_key": true,
      "api_dialect": "gemini",
      "default_model": "gemini-2.5-flash",
      "cheap_model": "gemini-2.0-flash",
      "aliases": [
        "google"
      ],
      "tags": [
        "code-execution",
        "frontier",
        "function-calling",
        "long-context",
        "multimodal",
        "streaming",
        "vision",
        "web-search"
      ],
      "models": [
        {
          "id": "flash-25",
          "model": "gemini-2.5-flash",
          "context_window_tokens": 1048576,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": true,
            "json_mode": "schema"
          }
        },
        {
          "id": "flash-20",
          "model": "gemini-2.0-flash",
          "context_window_tokens": 1048576,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": true
          }
        }
      ]
    },
    {
      "id": "deepseek",
      "name": "DeepSeek",
      "kind": "cloud",
      "description": "DeepSeek Chat and Coder models.",
      "env_var": "DEEPSEEK_API_KEY",
      "requires_key": true,
      "api_dialect": "openai-chat",
      "default_model": "deepseek-chat",
      "cheap_model": "deepseek-chat",
      "aliases": [
        "deep-seek"
      ],
      "tags": [
        "budget",
        "chinese",
        "code",
        "function-calling",
        "open-source",
        "reasoning",
        "streaming",
        "structured-output"
      ],
      "models": [
        {
          "id": "chat",
          "model": "deepseek-chat",
          "context_window_tokens": 64000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": false,
            "json_mode": "object"
          }
        }
      ]
    },
    {
      "id": "xai",
      "name": "xAI Grok",
      "kind": "cloud",
      "description": "Grok models (Grok-3, Grok-4).",
      "env_var": "XAI_API_KEY",
      "requires_key": true,
      "api_dialect": "openai-chat",
      "default_model": "grok-3",
      "cheap_model": "grok-3-mini-fast",
      "aliases": [
        "grok"
      ],
      "tags": [
        "frontier",
        "function-calling",
        "reasoning",
        "streaming",
        "structured-output",
        "vision",
        "web-search"
      ],
      "models": [
        {
          "id": "grok-3",
          "model": "grok-3",
          "context_window_tokens": 131072,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": false,
            "json_mode": "schema"
          }
        },
        {
          "id": "mini",
          "model": "grok-3-mini-fast",
          "context_window_tokens": 131072,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": true,
            "vision": false,
            "json_mode": "schema"
          }
        }
      ]
    },
    {
      "id": "groq",
      "name": "Groq",
      "kind": "cloud",
      "description": "Fast inference with Llama, Mixtral models.",
      "env_var": "GROQ_API_KEY",
      "requires_key": true,
      "api_dialect": "openai-chat",
      "default_model": "llama-3.3-70b-versatile",
      "cheap_model": "llama-3.1-8b-instant",
      "aliases": [],
      "tags": [
        "fast",
        "function-calling",
        "parallel-tools",
        "streaming",
        "structured-output"
      ],
      "models": [
        {
          "id": "llama-70b",
          "model": "llama-3.3-70b-versatile",
          "context_window_tokens": 128000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": false,
            "json_mode": "schema"
          }
        },
        {
          "id": "llama-8b",
          "model": "llama-3.1-8b-instant",
          "context_window_tokens": 128000,
          "max_output_tokens": 8192,
          "capabilities": {
            "reasoning": false,
            "vision": false,
            "json_mode": "schema"
          }
        }
      ]
    },
    {
      "id": "openrouter",
      "name": "OpenRouter",
      "kind": "cloud",
      "description": "200+ models via unified gateway.",
      "env_var": "OPENROUTER_API_KEY",
      "requires_key": true,
      "api_dialect": "openai-chat",
      "default_model": "anthropic/claude-sonnet-4-20250514",
      "cheap_model": "anthropic/claude-haiku-4-5-20251001",
      "aliases": [
        "or"
      ],
      "tags": [
        "function-calling",
        "streaming"
      ],
      "models": [
        {
          "id": "default",
          "model": "anthropic/claude-sonnet-4-20250514",
          "context_window_tokens": 128000,
          "max_output_tokens": 4096,
          "capabilities": {
            "reasoning": true,
            "vision": true,
            "json_mode": "schema"
          }
        },
        {
          "id": "cheap",
          "model": "anthropic/claude-haiku-4-5-20251001",
          "context_window_tokens": 128000,
          "max_output_tokens": 4096,
          "capabilities": {
            "reasoning": true,
            "vision": true,
            "json_mode": "schema"
          }
        }
      ]
    },
    {
      "id": "huggingface",
      "name": "Hugging Face",
      "kind": "cloud",
      "description": "Inference Providers router · 100+ open-weight models across 18 providers (Groq · Cerebras · Together · Scaleway · OVHcloud · …) · zero markup · :provider or :fastest/:cheapest routing suffix.",
      "env_var": "HF_TOKEN",
      "requires_key": true,
      "api_dialect": "openai-chat",
      "default_model": "Qwen/Qwen3.5-397B-A17B:fastest",
      "cheap_model": "Qwen/Qwen3.5-9B:cheapest",
      "aliases": [
        "hf"
      ],
      "tags": [
        "function-calling",
        "open-source",
        "streaming"
      ],
      "models": [
        {
          "id": "default",
          "model": "Qwen/Qwen3.5-397B-A17B:fastest",
          "context_window_tokens": 262144,
          "max_output_tokens": 32768,
          "capabilities": {
            "reasoning": false,
            "vision": true
          }
        },
        {
          "id": "cheap",
          "model": "Qwen/Qwen3.5-9B:cheapest",
          "context_window_tokens": 262144,
          "max_output_tokens": 16384,
          "capabilities": {
            "reasoning": false,
            "vision": true
          }
        }
      ]
    },
    {
      "id": "nvidia",
      "name": "NVIDIA",
      "kind": "cloud",
      "description": "NVIDIA API (integrate.api.nvidia.com) · Nemotron 3 family (Open Model License · agentic-first) + hosted open models · self-hosted NIM containers expose the same surface.",
      "env_var": "NVIDIA_API_KEY",
      "requires_key": true,
      "api_dialect": "openai-chat",
      "default_model": "nvidia/nemotron-3-super-120b-a12b",
      "cheap_model": "nvidia/nemotron-3-nano-30b-a3b",
      "aliases": [
        "nim",
        "nvidia-nim"
      ],
      "tags": [
        "function-calling",
        "open-source",
        "streaming"
      ],
      "models": [
        {
          "id": "default",
          "model": "nvidia/nemotron-3-super-120b-a12b",
          "context_window_tokens": 262144,
          "max_output_tokens": 32768,
          "capabilities": {
            "reasoning": false,
            "vision": false,
            "json_mode": "schema"
          }
        },
        {
          "id": "cheap",
          "model": "nvidia/nemotron-3-nano-30b-a3b",
          "context_window_tokens": 262144,
          "max_output_tokens": 16384,
          "capabilities": {
            "reasoning": false,
            "vision": false,
            "json_mode": "schema"
          }
        },
        {
          "id": "llama-70b",
          "model": "meta/llama-3.1-70b-instruct",
          "context_window_tokens": 128000,
          "max_output_tokens": 4096,
          "capabilities": {
            "reasoning": false,
            "vision": false,
            "json_mode": "schema"
          }
        }
      ]
    },
    {
      "id": "mock",
      "name": "Mock",
      "kind": "test",
      "description": "Deterministic test responses — no API calls, no keys needed.",
      "requires_key": false,
      "api_dialect": "mock",
      "default_model": "mock-default",
      "cheap_model": "mock-default",
      "aliases": [],
      "tags": [],
      "models": [
        {
          "id": "default",
          "model": "mock-default",
          "context_window_tokens": 1000000,
          "max_output_tokens": 1000000,
          "capabilities": {
            "reasoning": true,
            "vision": true,
            "json_mode": "schema"
          }
        }
      ]
    }
  ]
}