{
  "generatedAt": "2026-07-24T16:33:59.048Z",
  "frameworkVersion": "2.0.0",
  "evaluationStatusLabels": {
    "reported": "Author reported",
    "independently_verified": "Independently verified",
    "reproduced": "Reproduced",
    "no_shared_protocol": "No shared protocol",
    "not_evaluable": "Not evaluable",
    "deprecated": "Deprecated"
  },
  "sourceTypeLabels": {
    "author_reported": "Author reported",
    "independent": "Independent source",
    "reproduced": "Reproduced run"
  },
  "familyReadiness": [
    {
      "familySlug": "model-based-rl",
      "protocolReadiness": "public",
      "candidateScope": "Atari 100K and DMControl result intake",
      "nextAction": "Run the open-model harness with recorded seeds, environments, hardware, checkpoints, and raw logs."
    },
    {
      "familySlug": "driving-simulation",
      "protocolReadiness": "draft",
      "candidateScope": "Closed-loop scenario evaluation with fixed sensor configuration",
      "nextAction": "Publish a scenario suite, sensor schema, simulator version, and safety metrics before accepting results."
    },
    {
      "familySlug": "interactive-video-worlds",
      "protocolReadiness": "draft",
      "candidateScope": "Action-conditioned video rollouts under a fixed control interface",
      "nextAction": "Specify tasks, action space, latency measurement, rollout horizon, and evaluation capture procedure."
    },
    {
      "familySlug": "spatial-3d-worlds",
      "protocolReadiness": "draft",
      "candidateScope": "Navigable scene generation and geometry consistency",
      "nextAction": "Fix scene inputs, navigation tasks, reconstruction metrics, and export validation before accepting results."
    },
    {
      "familySlug": "latent-predictive-models",
      "protocolReadiness": "draft",
      "candidateScope": "Predictive representation transfer and physical reasoning",
      "nextAction": "Define downstream tasks, frozen-versus-finetuned settings, data splits, and transfer metrics."
    },
    {
      "familySlug": "embodied-robotics",
      "protocolReadiness": "draft",
      "candidateScope": "Embodied task success under a fixed robot and observation setup",
      "nextAction": "Fix robot embodiment, task set, observation mode, simulator or hardware version, and safety reporting."
    }
  ],
  "families": [
    {
      "slug": "driving-simulation",
      "title": "Driving Simulation World Models",
      "description": "Models evaluated for controllable, closed-loop autonomous-driving simulation. Metrics are meaningful only within the same sensor and scenario protocol.",
      "metrics": [
        "Scenario control",
        "Closed-loop realism",
        "Sensor fidelity"
      ],
      "modelSlugs": [
        "gaia-1",
        "gaia-2",
        "waabi-world",
        "copilot4d",
        "mile"
      ],
      "route": "/en/benchmarks/driving-simulation/"
    },
    {
      "slug": "interactive-video-worlds",
      "title": "Interactive Video World Models",
      "description": "Action-conditioned video world models evaluated on responsiveness, control following, rollout stability, and spatial persistence.",
      "metrics": [
        "Latency / FPS",
        "Action alignment",
        "Consistency horizon"
      ],
      "modelSlugs": [
        "wham",
        "wham-rt",
        "odyssey-2",
        "relic",
        "matrix-game-2",
        "oasis",
        "genie-3",
        "nvidia-cosmos",
        "genie-2",
        "unisim",
        "sora",
        "pandora",
        "genie-1",
        "gamengen",
        "emu-video",
        "stable-video-diffusion",
        "steve-1",
        "gen-3-alpha",
        "pixverse-r1"
      ],
      "route": "/en/benchmarks/interactive-video-worlds/"
    },
    {
      "slug": "spatial-3d-worlds",
      "title": "Spatial 3D World Models",
      "description": "Systems that create or reconstruct navigable 3D worlds. Geometry and export quality should not be compared to video-only metrics.",
      "metrics": [
        "Geometric consistency",
        "Navigability",
        "3D export"
      ],
      "modelSlugs": [
        "hy-world-2",
        "marble"
      ],
      "route": "/en/benchmarks/spatial-3d-worlds/"
    },
    {
      "slug": "model-based-rl",
      "title": "Model-Based RL",
      "description": "World models evaluated through downstream control and sample efficiency on shared reinforcement-learning benchmarks.",
      "metrics": [
        "Task return",
        "Sample efficiency",
        "Robustness"
      ],
      "modelSlugs": [
        "dreamer-v3",
        "dreamer-v2",
        "planet",
        "muzero",
        "td-mpc2",
        "diamond",
        "predictron",
        "ha-world-model",
        "iris",
        "imagination-augmented-agents",
        "value-prediction-network"
      ],
      "route": "/en/benchmarks/model-based-rl/"
    },
    {
      "slug": "latent-predictive-models",
      "title": "Latent Predictive World Models",
      "description": "Predictive representation models evaluated through physical reasoning, planning transfer, and training efficiency.",
      "metrics": [
        "Physical reasoning",
        "Planning transfer",
        "Training footprint"
      ],
      "modelSlugs": [
        "v-jepa",
        "v-jepa-2",
        "leworldmodel",
        "i-jepa",
        "rssm",
        "lwm"
      ],
      "route": "/en/benchmarks/latent-predictive-models/"
    },
    {
      "slug": "embodied-robotics",
      "title": "Embodied Robotics World Models",
      "description": "World models used for robot perception, control, or learned simulation. Results are comparable only within a shared robot task and protocol.",
      "metrics": [
        "Task success",
        "Transfer",
        "Physical consistency"
      ],
      "modelSlugs": [
        "ami-world-model",
        "3d-vla",
        "rt-2",
        "1x-world-model",
        "playworld"
      ],
      "route": "/en/benchmarks/embodied-robotics/"
    }
  ],
  "protocols": [
    {
      "slug": "atari-100k",
      "familySlug": "model-based-rl",
      "version": "v1",
      "title": "Atari 100K",
      "metrics": [
        "Mean HNS",
        "Sample budget"
      ],
      "sourceBenchmarkNames": [
        "Atari 100K"
      ],
      "resultMetric": "Mean HNS",
      "canonicalUnit": "x human",
      "harnessProtocolId": "model-based-rl/atari-100k/v1",
      "comparabilityNote": "Results require a documented task set, environment and ROM version, observation pipeline, seed count, interaction budget, Mean HNS, and human-normalized unit before they are comparable."
    },
    {
      "slug": "dmcontrol",
      "familySlug": "model-based-rl",
      "version": "v1",
      "title": "DeepMind Control Suite",
      "metrics": [
        "Mean return",
        "Environment set"
      ],
      "sourceBenchmarkNames": [
        "DMControl"
      ],
      "resultMetric": "Mean Return",
      "canonicalUnit": "avg return",
      "harnessProtocolId": "model-based-rl/dmcontrol/v1",
      "comparabilityNote": "Results require a documented DMControl task set, environment version, observation settings, seed count, and mean-return metric before they are comparable."
    }
  ],
  "results": [
    {
      "id": "dreamer-v3-atari-100k-0",
      "modelSlug": "dreamer-v3",
      "sourceBenchmarkName": "Atari 100K",
      "metric": "Mean HNS",
      "score": "Superhuman on 50+ games",
      "numericValue": 2.01,
      "unit": "x human",
      "sourceUrl": "https://arxiv.org/abs/2301.04104",
      "recordedAt": "2026-07-24",
      "protocolSlug": "atari-100k",
      "protocolVersion": "v1",
      "status": "reported",
      "sourceType": "author_reported",
      "comparisonEligibility": "evidence_only",
      "comparisonExclusionReason": "The source record does not yet document the task set, environment configuration, observation pipeline, and seed count required for a valid cross-model comparison."
    },
    {
      "id": "dreamer-v3-dmcontrol-1",
      "modelSlug": "dreamer-v3",
      "sourceBenchmarkName": "DMControl",
      "metric": "Mean Return",
      "score": "State-of-the-art",
      "numericValue": 901,
      "unit": "avg return",
      "sourceUrl": "https://arxiv.org/abs/2301.04104",
      "recordedAt": "2026-07-24",
      "protocolSlug": "dmcontrol",
      "protocolVersion": "v1",
      "status": "reported",
      "sourceType": "author_reported",
      "comparisonEligibility": "evidence_only",
      "comparisonExclusionReason": "The source record does not yet document the task set, environment configuration, observation pipeline, and seed count required for a valid cross-model comparison."
    },
    {
      "id": "iris-atari-100k-0",
      "modelSlug": "iris",
      "sourceBenchmarkName": "Atari 100K",
      "metric": "Mean HNS",
      "score": "Strong performance",
      "numericValue": 1.046,
      "unit": "x human",
      "sourceUrl": "https://arxiv.org/abs/2209.00588",
      "recordedAt": "2026-02-25",
      "protocolSlug": "atari-100k",
      "protocolVersion": "v1",
      "status": "reported",
      "sourceType": "author_reported",
      "comparisonEligibility": "evidence_only",
      "comparisonExclusionReason": "The source record does not yet document the task set, environment configuration, observation pipeline, and seed count required for a valid cross-model comparison."
    },
    {
      "id": "diamond-atari-100k-0",
      "modelSlug": "diamond",
      "sourceBenchmarkName": "Atari 100K",
      "metric": "Mean HNS",
      "score": "State-of-the-art",
      "numericValue": 1.56,
      "unit": "x human",
      "sourceUrl": "https://arxiv.org/abs/2405.12399",
      "recordedAt": "2026-03-10",
      "protocolSlug": "atari-100k",
      "protocolVersion": "v1",
      "status": "reported",
      "sourceType": "author_reported",
      "comparisonEligibility": "evidence_only",
      "comparisonExclusionReason": "The source record does not yet document the task set, environment configuration, observation pipeline, and seed count required for a valid cross-model comparison."
    }
  ],
  "reportedEvidence": [
    {
      "id": "dreamer-v3-evidence-0",
      "modelSlug": "dreamer-v3",
      "sourceBenchmarkName": "Atari 100K",
      "metric": "Mean HNS",
      "score": "Superhuman on 50+ games",
      "numericValue": 2.01,
      "unit": "x human",
      "sourceUrl": "https://arxiv.org/abs/2301.04104",
      "recordedAt": "2026-07-24",
      "protocolSlug": "atari-100k"
    },
    {
      "id": "dreamer-v3-evidence-1",
      "modelSlug": "dreamer-v3",
      "sourceBenchmarkName": "DMControl",
      "metric": "Mean Return",
      "score": "State-of-the-art",
      "numericValue": 901,
      "unit": "avg return",
      "sourceUrl": "https://arxiv.org/abs/2301.04104",
      "recordedAt": "2026-07-24",
      "protocolSlug": "dmcontrol"
    },
    {
      "id": "dreamer-v2-evidence-0",
      "modelSlug": "dreamer-v2",
      "sourceBenchmarkName": "Atari 200M",
      "metric": "Mean HNS",
      "score": "Human-level (first model-based)",
      "numericValue": 1,
      "unit": "x human",
      "sourceUrl": "https://arxiv.org/abs/2010.02193",
      "recordedAt": "2026-02-18",
      "exclusionReason": "Source benchmark name does not match a registered public protocol."
    },
    {
      "id": "planet-evidence-0",
      "modelSlug": "planet",
      "sourceBenchmarkName": "DMControl Suite",
      "metric": "Sample Efficiency",
      "score": "Competitive at 10x less data",
      "numericValue": 10,
      "unit": "x more efficient",
      "sourceUrl": "https://arxiv.org/abs/1811.04551",
      "recordedAt": "2026-02-05",
      "exclusionReason": "Source benchmark name does not match a registered public protocol."
    },
    {
      "id": "muzero-evidence-2",
      "modelSlug": "muzero",
      "sourceBenchmarkName": "Atari",
      "metric": "Mean HNS",
      "score": "State-of-the-art",
      "numericValue": 7.31,
      "unit": "x human",
      "sourceUrl": "https://arxiv.org/abs/1911.08265",
      "recordedAt": "2026-02-20",
      "exclusionReason": "Source benchmark name does not match a registered public protocol."
    },
    {
      "id": "td-mpc2-evidence-0",
      "modelSlug": "td-mpc2",
      "sourceBenchmarkName": "DMControl (30 tasks)",
      "metric": "Mean Return",
      "score": "State-of-the-art",
      "numericValue": 879,
      "unit": "avg return",
      "sourceUrl": "https://arxiv.org/abs/2310.16828",
      "recordedAt": "2026-03-05",
      "exclusionReason": "Source benchmark name does not match a registered public protocol."
    },
    {
      "id": "iris-evidence-0",
      "modelSlug": "iris",
      "sourceBenchmarkName": "Atari 100K",
      "metric": "Mean HNS",
      "score": "Strong performance",
      "numericValue": 1.046,
      "unit": "x human",
      "sourceUrl": "https://arxiv.org/abs/2209.00588",
      "recordedAt": "2026-02-25",
      "protocolSlug": "atari-100k"
    },
    {
      "id": "imagination-augmented-agents-evidence-1",
      "modelSlug": "imagination-augmented-agents",
      "sourceBenchmarkName": "Atari",
      "metric": "Mean HNS",
      "score": "Improved over baselines",
      "numericValue": 1.15,
      "unit": "x baseline",
      "sourceUrl": "https://arxiv.org/abs/1707.06203",
      "recordedAt": "2026-01-20",
      "exclusionReason": "Source benchmark name does not match a registered public protocol."
    },
    {
      "id": "value-prediction-network-evidence-0",
      "modelSlug": "value-prediction-network",
      "sourceBenchmarkName": "Atari",
      "metric": "Mean HNS",
      "score": "Improved over model-free baselines",
      "numericValue": 1.08,
      "unit": "x baseline",
      "sourceUrl": "https://arxiv.org/abs/1707.03497",
      "recordedAt": "2026-01-15",
      "exclusionReason": "Source benchmark name does not match a registered public protocol."
    },
    {
      "id": "diamond-evidence-0",
      "modelSlug": "diamond",
      "sourceBenchmarkName": "Atari 100K",
      "metric": "Mean HNS",
      "score": "State-of-the-art",
      "numericValue": 1.56,
      "unit": "x human",
      "sourceUrl": "https://arxiv.org/abs/2405.12399",
      "recordedAt": "2026-03-10",
      "protocolSlug": "atari-100k"
    }
  ],
  "comparisonGroups": [
    {
      "protocolSlug": "atari-100k",
      "resultCount": 3,
      "comparisonEligible": false
    },
    {
      "protocolSlug": "dmcontrol",
      "resultCount": 1,
      "comparisonEligible": false
    }
  ]
}