{
  "version": 1,
  "status": "illustrative-game-inputs-not-verified-benchmarks",
  "provenance": "Retained from the original Model Golf game design. The underlying run artifacts, ticket texts, labels and per-ticket costs have not been verified. This export documents the game inputs only.",
  "tickets": 10,
  "cost_unit": "USD per 1000 tickets (illustrative)",
  "calculation": "For each model, add c * remaining_tickets / 10, then remove remaining tickets whose r position is 1. Sum all model costs.",
  "assumptions": [
    "Perfect detection of incorrect answers",
    "Average listed cost applies to every ticket",
    "No checking, retry, routing or latency costs"
  ],
  "fields": {
    "id": "Game identifier",
    "model": "Model label",
    "provider": "Maker label",
    "setting": "Configuration label",
    "s": "Pass count in game scorecard",
    "r": "Ten positions: 1 passes, 0 fails",
    "c": "Illustrative USD per 1000 tickets",
    "size": "Parameter-count label, where provided"
  },
  "setups": [
    {
      "id": "l",
      "model": "Claude Haiku 5.5",
      "provider": "Anthropic",
      "setting": "default",
      "s": 10,
      "r": "1111111111",
      "c": 0.03948
    },
    {
      "id": "k",
      "model": "Claude Sonnet 5.5",
      "provider": "Anthropic",
      "setting": "default",
      "s": 10,
      "r": "1111111111",
      "c": 0.6886
    },
    {
      "id": "h",
      "model": "Claude Fable 5.1",
      "provider": "Anthropic",
      "setting": "default",
      "s": 10,
      "r": "1111111111",
      "c": 3.068
    },
    {
      "id": "a",
      "model": "Claude Opus 5.5",
      "provider": "Anthropic",
      "setting": "default",
      "s": 10,
      "r": "1111111111",
      "c": 1.5132
    },
    {
      "id": "m",
      "model": "Gemini 2.5 Flash-Lite",
      "provider": "Google",
      "setting": "default",
      "s": 8,
      "r": "0111111110",
      "c": 0.01471
    },
    {
      "id": "b",
      "model": "Gemini 3.8 Flash",
      "provider": "Google",
      "setting": "default",
      "s": 10,
      "r": "1111111111",
      "c": 0.73508
    },
    {
      "id": "p",
      "model": "Mercury 2.5",
      "provider": "Inception",
      "setting": "default",
      "s": 9,
      "r": "0111111111",
      "c": 0.09691
    },
    {
      "id": "o",
      "model": "Llama 3.1 8B Instruct",
      "provider": "Meta",
      "setting": "default",
      "s": 9,
      "r": "0111111111",
      "c": 0.02996,
      "size": "8B"
    },
    {
      "id": "n",
      "model": "Llama 3.3 70B Instruct",
      "provider": "Meta",
      "setting": "default",
      "s": 9,
      "r": "0111111111",
      "c": 0.11174,
      "size": "70B"
    },
    {
      "id": "e",
      "model": "Mistral Nemo 12B",
      "provider": "Mistral",
      "setting": "default",
      "s": 9,
      "r": "0111111111",
      "c": 0.00371,
      "size": "12B"
    },
    {
      "id": "d",
      "model": "Ministral 8B",
      "provider": "Mistral",
      "setting": "default",
      "s": 10,
      "r": "1111111111",
      "c": 0.02031,
      "size": "8B"
    },
    {
      "id": "q",
      "model": "gpt-oss-20b",
      "provider": "OpenAI",
      "setting": "default",
      "s": 9,
      "r": "0111111111",
      "c": 0.01507,
      "size": "21B"
    },
    {
      "id": "f",
      "model": "GPT-6 Luna",
      "provider": "OpenAI",
      "setting": "default",
      "s": 10,
      "r": "1111111111",
      "c": 0.0269
    },
    {
      "id": "g",
      "model": "GPT-5.6 Terra",
      "provider": "OpenAI",
      "setting": "low reasoning",
      "s": 10,
      "r": "1111111111",
      "c": 0.3698
    },
    {
      "id": "c",
      "model": "GPT-5.6 Terra",
      "provider": "OpenAI",
      "setting": "high reasoning",
      "s": 10,
      "r": "1111111111",
      "c": 0.413
    },
    {
      "id": "i",
      "model": "GPT-5.6 Sol",
      "provider": "OpenAI",
      "setting": "default",
      "s": 10,
      "r": "1111111111",
      "c": 0.716
    },
    {
      "id": "j",
      "model": "GPT-5.6 Sol",
      "provider": "OpenAI",
      "setting": "high reasoning",
      "s": 10,
      "r": "1111111111",
      "c": 0.786
    }
  ]
}