{
  "models": [
    {
      "id": "m0",
      "name": "Claude Opus 5 (high)",
      "provider": "Anthropic",
      "color": "#CC785C",
      "before": 0.6210604709915126,
      "after": 0.8956726734441848,
      "logo": "assets/model-logos/anthropic.svg"
    },
    {
      "id": "m1",
      "name": "Grok 4.6 (high)",
      "provider": "xAI",
      "color": "#736CD3",
      "before": 0.5817587217671679,
      "after": 0.8906746770585875,
      "logo": "assets/model-logos/xai.svg"
    },
    {
      "id": "m2",
      "name": "GPT-6 Astra (high)",
      "provider": "OpenAI",
      "color": "#1F1F1F",
      "before": 0.5750502479341251,
      "after": 0.8805690502154662,
      "logo": "assets/model-logos/openai.svg"
    },
    {
      "id": "m3",
      "name": "GPT-5.6 Sol (high)",
      "provider": "OpenAI",
      "color": "#1F1F1F",
      "before": 0.45931067976335743,
      "after": 0.8771229489361815,
      "logo": "assets/model-logos/openai.svg"
    },
    {
      "id": "m4",
      "name": "GPT-5.6 Terra (high)",
      "provider": "OpenAI",
      "color": "#1F1F1F",
      "before": 0.3786913833449128,
      "after": 0.8616368669762594,
      "logo": "assets/model-logos/openai.svg"
    },
    {
      "id": "m5",
      "name": "GPT-5.6 Luna (high)",
      "provider": "OpenAI",
      "color": "#1F1F1F",
      "before": 0.3488996168821823,
      "after": 0.8733657264304917,
      "logo": "assets/model-logos/openai.svg"
    },
    {
      "id": "m6",
      "name": "MiniMax-M2.5",
      "provider": "MiniMax",
      "color": "#EB3568",
      "before": 0.16844029210192132,
      "after": 0.7808273762995732,
      "logo": "assets/model-logos/minimax.svg"
    },
    {
      "id": "m7",
      "name": "Kimi K2.5",
      "provider": "Moonshot",
      "color": "#047AFE",
      "before": 0.0893298488149468,
      "after": 0.8596151081556477,
      "logo": "assets/model-logos/kimi.jpg"
    },
    {
      "id": "m8",
      "name": "GLM-5",
      "provider": "Z.ai",
      "color": "#1C7FF8",
      "before": 0.05191218689557857,
      "after": 0.8020923994083032,
      "logo": "assets/model-logos/zai.svg"
    },
    {
      "id": "m9",
      "name": "Qwen3 235B 2507",
      "provider": "Alibaba",
      "color": "#FF7018",
      "before": -0.030246655888524755,
      "after": 0.766647178883721,
      "logo": "assets/model-logos/alibaba.svg"
    },
    {
      "id": "m10",
      "name": "DeepSeek V3.2",
      "provider": "DeepSeek",
      "color": "#2243E6",
      "before": -0.05420376313519638,
      "after": 0.8619109126822841,
      "logo": "assets/model-logos/deepseek.svg"
    },
    {
      "id": "m11",
      "name": "Nemotron 3 Super",
      "provider": "NVIDIA",
      "color": "#76B900",
      "before": -0.23293819782826322,
      "after": 0.8003240774388909,
      "logo": "assets/model-logos/nvidia.svg"
    },
    {
      "id": "m12",
      "name": "Mistral Large 3",
      "provider": "Mistral",
      "color": "#F32E02",
      "before": -0.3620032386340157,
      "after": 0.6383063753018331,
      "logo": "assets/model-logos/mistral.svg"
    }
  ],
  "source": {
    "title": "Language Models as AI Research World Models",
    "url": "https://arxiv.org/abs/2610.12235",
    "version": "v2",
    "figure": 3,
    "metric": "Spearman \u03c1",
    "environment": "OLMo3-100M",
    "queries": 160,
    "records": 320,
    "note": "Endpoint measurements; animation interpolates between conditions. Rank uses unrounded values."
  },
  "efforts": [
    {
      "label": "LOW",
      "w0": 0.6091314182884019,
      "knowledge": 0.8921454567041198
    },
    {
      "label": "MEDIUM",
      "w0": 0.5853497575964278,
      "knowledge": 0.898057706318918
    },
    {
      "label": "HIGH",
      "w0": 0.6210604709915126,
      "knowledge": 0.8956726734441848
    },
    {
      "label": "XHIGH",
      "w0": 0.6540134816755361,
      "knowledge": 0.908717193038081
    },
    {
      "label": "MAX",
      "w0": 0.6476407140543134,
      "knowledge": 0.9041222549924585
    }
  ]
}
