{
  "models": [
    {
      "provider": "anthropic",
      "providerLabel": "Anthropic",
      "name": "Claude Opus 4.6",
      "id": "claude-opus-4-6",
      "tier": "frontier",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "inputPrice": 5.00,
      "outputPrice": 25.00,
      "released": "2026-02-05",
      "gpqa": 91.3,
      "swebench": 80.8,
      "capabilities": ["vision", "tool-use", "reasoning", "code", "web-search", "computer-use"],
      "notes": "Strongest reasoning. 14.5-hour autonomous task horizon.",
      "source": "https://docs.anthropic.com/en/docs/about-claude/models"
    },
    {
      "provider": "anthropic",
      "providerLabel": "Anthropic",
      "name": "Claude Sonnet 4.6",
      "id": "claude-sonnet-4-6",
      "tier": "performance",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "inputPrice": 3.00,
      "outputPrice": 15.00,
      "released": "2026-02-17",
      "gpqa": 74.1,
      "swebench": 79.6,
      "capabilities": ["vision", "tool-use", "reasoning", "code", "web-search", "computer-use"],
      "notes": "Best value for complex tasks. Near-Opus coding at 60% cost.",
      "source": "https://docs.anthropic.com/en/docs/about-claude/models"
    },
    {
      "provider": "anthropic",
      "providerLabel": "Anthropic",
      "name": "Claude Haiku 4.5",
      "id": "claude-haiku-4-5",
      "tier": "speed",
      "contextWindow": 200000,
      "maxOutput": 64000,
      "inputPrice": 1.00,
      "outputPrice": 5.00,
      "released": "2025-10-15",
      "swebench": 73.3,
      "capabilities": ["vision", "tool-use", "code"],
      "notes": "Fast and cheap. Good for classification, extraction, routing.",
      "source": "https://docs.anthropic.com/en/docs/about-claude/models"
    },
    {
      "provider": "openai",
      "providerLabel": "OpenAI",
      "name": "GPT-5.4",
      "id": "gpt-5-4",
      "tier": "frontier",
      "contextWindow": 1000000,
      "maxOutput": 128000,
      "inputPrice": 2.50,
      "outputPrice": 20.00,
      "released": "2026-03-05",
      "capabilities": ["vision", "reasoning", "tool-use", "code", "computer-use"],
      "notes": "OpenAI flagship. Native computer use. Successor to GPT-5/5.2 line.",
      "source": "https://openai.com/api/pricing/"
    },
    {
      "provider": "openai",
      "providerLabel": "OpenAI",
      "name": "o3",
      "id": "o3",
      "tier": "frontier",
      "contextWindow": 200000,
      "maxOutput": 100000,
      "inputPrice": 2.00,
      "outputPrice": 8.00,
      "released": "2025-04-16",
      "gpqa": 87.7,
      "capabilities": ["reasoning", "tool-use", "vision"],
      "notes": "Chain-of-thought reasoning. Hidden thinking tokens billed as output.",
      "source": "https://openai.com/api/pricing/"
    },
    {
      "provider": "openai",
      "providerLabel": "OpenAI",
      "name": "GPT-4o",
      "id": "gpt-4o",
      "tier": "performance",
      "contextWindow": 128000,
      "maxOutput": 16000,
      "inputPrice": 2.50,
      "outputPrice": 10.00,
      "released": "2024-05-13",
      "mmlu": 88.7,
      "gpqa": 53.6,
      "humaneval": 90.2,
      "capabilities": ["vision", "tool-use", "code", "function-calling"],
      "notes": "Retiring 2026. Still widely deployed. Solid all-rounder.",
      "source": "https://openai.com/api/pricing/"
    },
    {
      "provider": "openai",
      "providerLabel": "OpenAI",
      "name": "GPT-4o-mini",
      "id": "gpt-4o-mini",
      "tier": "speed",
      "contextWindow": 128000,
      "maxOutput": 16000,
      "inputPrice": 0.15,
      "outputPrice": 0.60,
      "released": "2024-07-18",
      "mmlu": 82.0,
      "humaneval": 87.2,
      "capabilities": ["vision", "tool-use", "function-calling"],
      "notes": "Cheapest OpenAI option. Good for high-volume, simple tasks.",
      "source": "https://openai.com/api/pricing/"
    },
    {
      "provider": "openai",
      "providerLabel": "OpenAI",
      "name": "o3-mini",
      "id": "o3-mini",
      "tier": "performance",
      "contextWindow": 200000,
      "maxOutput": 100000,
      "inputPrice": 1.10,
      "outputPrice": 4.40,
      "released": "2025-01-31",
      "gpqa": 79.7,
      "mmlu": 81.1,
      "capabilities": ["reasoning", "tool-use"],
      "notes": "Budget reasoning model. Good for math, logic, structured analysis.",
      "source": "https://openai.com/api/pricing/"
    },
    {
      "provider": "google",
      "providerLabel": "Google",
      "name": "Gemini 3.1 Pro",
      "id": "gemini-3-1-pro",
      "tier": "frontier",
      "contextWindow": 1000000,
      "maxOutput": 64000,
      "inputPrice": 2.00,
      "outputPrice": 12.00,
      "released": "2026-02-19",
      "swebench": 80.6,
      "capabilities": ["vision", "reasoning", "tool-use", "code", "grounding"],
      "notes": "Google flagship. Tiered pricing: $4.00/$18.00 above 200K context.",
      "source": "https://ai.google.dev/gemini-api/docs/pricing"
    },
    {
      "provider": "google",
      "providerLabel": "Google",
      "name": "Gemini 2.5 Pro",
      "id": "gemini-2-5-pro",
      "tier": "performance",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "inputPrice": 1.25,
      "outputPrice": 10.00,
      "released": "2025-03-25",
      "gpqa": 84.0,
      "mmlu": 86.0,
      "capabilities": ["vision", "reasoning", "tool-use", "code", "grounding"],
      "notes": "Strong price/performance for long context. Tiered: $2.50/$15.00 above 200K.",
      "source": "https://ai.google.dev/gemini-api/docs/pricing"
    },
    {
      "provider": "google",
      "providerLabel": "Google",
      "name": "Gemini 2.5 Flash",
      "id": "gemini-2-5-flash",
      "tier": "speed",
      "contextWindow": 1000000,
      "maxOutput": 65536,
      "inputPrice": 0.30,
      "outputPrice": 2.50,
      "released": "2025-06-17",
      "capabilities": ["vision", "reasoning", "tool-use"],
      "notes": "Extremely cheap for 1M context. Good for summarization, extraction.",
      "source": "https://ai.google.dev/gemini-api/docs/pricing"
    },
    {
      "provider": "meta",
      "providerLabel": "Meta",
      "name": "Llama 4 Maverick",
      "id": "llama-4-maverick",
      "tier": "performance",
      "contextWindow": 1000000,
      "maxOutput": 16384,
      "inputPrice": 0.15,
      "outputPrice": 0.60,
      "released": "2025-04-05",
      "mmlu": 85.5,
      "gpqa": 69.8,
      "capabilities": ["vision", "tool-use", "code"],
      "notes": "Open weights. 400B MoE. Matches GPT-4o on multimodal benchmarks.",
      "source": "https://www.llama.com/models/llama-4/"
    },
    {
      "provider": "meta",
      "providerLabel": "Meta",
      "name": "Llama 4 Scout",
      "id": "llama-4-scout",
      "tier": "speed",
      "contextWindow": 10000000,
      "maxOutput": 16384,
      "inputPrice": 0.15,
      "outputPrice": 0.60,
      "released": "2025-04-05",
      "mmlu": 79.6,
      "gpqa": 57.2,
      "capabilities": ["vision", "tool-use"],
      "notes": "10M context window. Open weights. 95%+ retrieval accuracy up to 8M tokens.",
      "source": "https://www.llama.com/models/llama-4/"
    },
    {
      "provider": "deepseek",
      "providerLabel": "DeepSeek",
      "name": "DeepSeek V3.2",
      "id": "deepseek-v3-2",
      "tier": "speed",
      "contextWindow": 128000,
      "maxOutput": 16000,
      "inputPrice": 0.28,
      "outputPrice": 0.42,
      "released": "2026-01-15",
      "mmlu": 88.5,
      "gpqa": 59.1,
      "capabilities": ["reasoning", "code", "tool-use"],
      "notes": "Unified chat + reasoning. Cache hits at $0.028/MTok. Open weights.",
      "source": "https://api-docs.deepseek.com/quick_start/pricing"
    },
    {
      "provider": "xai",
      "providerLabel": "xAI",
      "name": "Grok 3",
      "id": "grok-3",
      "tier": "performance",
      "contextWindow": 131072,
      "maxOutput": 16000,
      "inputPrice": 3.00,
      "outputPrice": 15.00,
      "released": "2025-02-17",
      "capabilities": ["vision", "reasoning", "tool-use"],
      "notes": "RL-based reasoning. 10x compute vs predecessors. 128K context.",
      "source": "https://docs.x.ai/docs/models"
    },
    {
      "provider": "xai",
      "providerLabel": "xAI",
      "name": "Grok 4.1 Fast",
      "id": "grok-4-1-fast",
      "tier": "speed",
      "contextWindow": 2000000,
      "maxOutput": 16000,
      "inputPrice": 0.20,
      "outputPrice": 0.50,
      "released": "2025-07-15",
      "capabilities": ["vision", "reasoning", "tool-use"],
      "notes": "2M context. Budget xAI option for high-volume tasks.",
      "source": "https://docs.x.ai/docs/models"
    },
    {
      "provider": "mistral",
      "providerLabel": "Mistral",
      "name": "Mistral Large 3",
      "id": "mistral-large-3",
      "tier": "performance",
      "contextWindow": 262000,
      "maxOutput": 16000,
      "inputPrice": 0.50,
      "outputPrice": 1.50,
      "released": "2025-12-04",
      "capabilities": ["tool-use", "code", "multilingual", "reasoning"],
      "notes": "675B MoE. 6x cheaper than Large 2. Strong multilingual (12+ languages).",
      "source": "https://mistral.ai/products/pricing"
    }
  ],
  "lastUpdated": "2026-03-19"
}