{
  "catalogue_version": 4,
  "generated_at": "2026-09-25T23:55:57.618Z",
  "source": "gateway config/providers_catalogue.php (ids) + ProviderSeeder::catalogue() / shared/providers/providers.yaml (USD cents/1M → USD/1M) + catalogue/model-catalogue.src.json (pool/latency/benchmark) — runtime source of truth",
  "unit_cost": "usd_per_1m_tokens",
  "candidates": [
    {
      "provider": "anthropic",
      "model": "claude-opus-4-8",
      "cost_per_1m_input": 5,
      "cost_per_1m_output": 25,
      "is_local": false,
      "benchmark_quality": 0.94,
      "p50_latency_ms": 900
    },
    {
      "provider": "anthropic",
      "model": "claude-haiku-4-5",
      "cost_per_1m_input": 1,
      "cost_per_1m_output": 5,
      "is_local": false,
      "benchmark_quality": 0.78,
      "p50_latency_ms": 400
    },
    {
      "provider": "anthropic",
      "model": "claude-fable-5",
      "cost_per_1m_input": 10,
      "cost_per_1m_output": 50,
      "is_local": false,
      "benchmark_quality": 0.88,
      "p50_latency_ms": 1000
    },
    {
      "provider": "anthropic",
      "model": "claude-sonnet-4-6",
      "cost_per_1m_input": 3,
      "cost_per_1m_output": 15,
      "is_local": false,
      "benchmark_quality": 0.86,
      "p50_latency_ms": 800
    },
    {
      "provider": "anthropic",
      "model": "claude-opus-4-6",
      "cost_per_1m_input": 5,
      "cost_per_1m_output": 25,
      "is_local": false,
      "benchmark_quality": 0.92,
      "p50_latency_ms": 950
    },
    {
      "provider": "openai",
      "model": "gpt-5.5",
      "cost_per_1m_input": 5,
      "cost_per_1m_output": 30,
      "is_local": false,
      "benchmark_quality": 0.93,
      "p50_latency_ms": 900
    },
    {
      "provider": "openai",
      "model": "gpt-5.5-pro",
      "cost_per_1m_input": 30,
      "cost_per_1m_output": 180,
      "is_local": false,
      "benchmark_quality": 0.95,
      "p50_latency_ms": 1200
    },
    {
      "provider": "openai",
      "model": "gpt-5.4",
      "cost_per_1m_input": 2.5,
      "cost_per_1m_output": 15,
      "is_local": false,
      "benchmark_quality": 0.9,
      "p50_latency_ms": 700
    },
    {
      "provider": "openai",
      "model": "gpt-5.4-mini",
      "cost_per_1m_input": 0.75,
      "cost_per_1m_output": 4.5,
      "is_local": false,
      "benchmark_quality": 0.82,
      "p50_latency_ms": 450
    },
    {
      "provider": "openai",
      "model": "gpt-5.4-nano",
      "cost_per_1m_input": 0.2,
      "cost_per_1m_output": 1.25,
      "is_local": false,
      "benchmark_quality": 0.74,
      "p50_latency_ms": 350
    },
    {
      "provider": "openai",
      "model": "gpt-5.3-codex",
      "cost_per_1m_input": 1.75,
      "cost_per_1m_output": 14,
      "is_local": false,
      "benchmark_quality": 0.88,
      "p50_latency_ms": 600
    },
    {
      "provider": "openai",
      "model": "gpt-4o",
      "cost_per_1m_input": 2.5,
      "cost_per_1m_output": 10,
      "is_local": false,
      "benchmark_quality": 0.85,
      "p50_latency_ms": 600
    },
    {
      "provider": "openai",
      "model": "gpt-4o-mini",
      "cost_per_1m_input": 0.15,
      "cost_per_1m_output": 0.6,
      "is_local": false,
      "benchmark_quality": 0.76,
      "p50_latency_ms": 400
    },
    {
      "provider": "gemini",
      "model": "gemini-3.5-flash",
      "cost_per_1m_input": 1.5,
      "cost_per_1m_output": 9,
      "is_local": false,
      "benchmark_quality": 0.84,
      "p50_latency_ms": 500
    },
    {
      "provider": "gemini",
      "model": "gemini-3.1-flash-lite",
      "cost_per_1m_input": 0.25,
      "cost_per_1m_output": 1.5,
      "is_local": false,
      "benchmark_quality": 0.72,
      "p50_latency_ms": 350
    },
    {
      "provider": "gemini",
      "model": "gemini-3.1-pro-preview",
      "cost_per_1m_input": 2,
      "cost_per_1m_output": 12,
      "is_local": false,
      "benchmark_quality": 0.89,
      "p50_latency_ms": 800
    },
    {
      "provider": "gemini",
      "model": "gemini-2.5-pro",
      "cost_per_1m_input": 1.25,
      "cost_per_1m_output": 10,
      "is_local": false,
      "benchmark_quality": 0.87,
      "p50_latency_ms": 700
    },
    {
      "provider": "gemini",
      "model": "gemini-2.5-flash",
      "cost_per_1m_input": 0.3,
      "cost_per_1m_output": 2.5,
      "is_local": false,
      "benchmark_quality": 0.8,
      "p50_latency_ms": 450
    },
    {
      "provider": "gemini",
      "model": "gemini-2.5-flash-lite",
      "cost_per_1m_input": 0.1,
      "cost_per_1m_output": 0.4,
      "is_local": false,
      "benchmark_quality": 0.7,
      "p50_latency_ms": 350
    },
    {
      "provider": "grok",
      "model": "grok-4.3",
      "cost_per_1m_input": 1.25,
      "cost_per_1m_output": 2.5,
      "is_local": false,
      "benchmark_quality": 0.86,
      "p50_latency_ms": 700
    },
    {
      "provider": "grok",
      "model": "grok-build-0.1",
      "cost_per_1m_input": 1,
      "cost_per_1m_output": 2,
      "is_local": false,
      "benchmark_quality": 0.83,
      "p50_latency_ms": 550
    },
    {
      "provider": "deepseek",
      "model": "deepseek-v4-pro",
      "cost_per_1m_input": 0.435,
      "cost_per_1m_output": 0.87,
      "is_local": false,
      "benchmark_quality": 0.88,
      "p50_latency_ms": 800
    },
    {
      "provider": "deepseek",
      "model": "deepseek-v4-flash",
      "cost_per_1m_input": 0.14,
      "cost_per_1m_output": 0.28,
      "is_local": false,
      "benchmark_quality": 0.79,
      "p50_latency_ms": 450
    },
    {
      "provider": "deepseek",
      "model": "deepseek-chat",
      "cost_per_1m_input": 0.14,
      "cost_per_1m_output": 0.28,
      "is_local": false,
      "benchmark_quality": 0.78,
      "p50_latency_ms": 500
    }
  ],
  "benchmark_scores": {
    "anthropic:claude-opus-4-7": 0.97,
    "anthropic:claude-opus-4-8": 0.94,
    "anthropic:claude-sonnet-4-6": 0.86,
    "anthropic:claude-haiku-4-5": 0.78,
    "anthropic:claude-3-5-sonnet": 0.92,
    "anthropic:claude-3-5-haiku": 0.84,
    "openai:gpt-5": 0.96,
    "openai:gpt-5.5": 0.93,
    "openai:gpt-5.4": 0.9,
    "openai:gpt-5-mini": 0.88,
    "openai:gpt-5.4-mini": 0.82,
    "openai:gpt-4o": 0.85,
    "openai:gpt-4o-mini": 0.76,
    "openai:gpt-4": 0.91,
    "openai:o1": 0.94,
    "openai:o3": 0.95,
    "grok:grok-4-pro": 0.94,
    "grok:grok-4": 0.9,
    "grok:grok-4.3": 0.86,
    "grok:grok-4-mini": 0.83,
    "google:gemini-2.5-pro": 0.93,
    "google:gemini-2.5-flash": 0.86,
    "gemini:gemini-2.5-pro": 0.87,
    "gemini:gemini-2.5-flash": 0.8,
    "gemini:gemini-3.5-flash": 0.84,
    "ollama:qwen2.5:72b": 0.89,
    "ollama:qwen2.5:32b": 0.86,
    "ollama:qwen2.5:14b": 0.82,
    "ollama:qwen2.5:7b": 0.76,
    "ollama:deepseek-v3": 0.9,
    "ollama:deepseek-r1": 0.91,
    "ollama:llama3.3:70b": 0.88,
    "ollama:llama3.1:70b": 0.86,
    "ollama:llama3.1:8b": 0.75,
    "ollama:mistral-large": 0.88,
    "ollama:mistral:7b": 0.73,
    "ollama:phi-3.5:3.8b": 0.72,
    "ollama:phi-4:14b": 0.83,
    "ollama:gemma2:27b": 0.87,
    "ollama:gemma2:9b": 0.79,
    "mistral:mistral-large-2": 0.89,
    "mistral:mistral-small-3": 0.78,
    "anthropic:claude-fable-5": 0.88,
    "anthropic:claude-opus-4-6": 0.92,
    "openai:gpt-5.5-pro": 0.95,
    "openai:gpt-5.4-nano": 0.74,
    "openai:gpt-5.3-codex": 0.88,
    "gemini:gemini-3.1-flash-lite": 0.72,
    "gemini:gemini-3.1-pro-preview": 0.89,
    "gemini:gemini-2.5-flash-lite": 0.7,
    "grok:grok-build-0.1": 0.83,
    "deepseek:deepseek-v4-pro": 0.88,
    "deepseek:deepseek-v4-flash": 0.79,
    "deepseek:deepseek-chat": 0.78
  }
}