{
  "schemaVersion": 1,
  "benchmark": {
    "id": "videophy-2",
    "name": "VideoPhy",
    "version": "2 · human evaluation",
    "category": "video",
    "question": "Does the action obey basic physics?",
    "summary": "Human reviewers check whether a generated video both follows the prompt and respects physical commonsense.",
    "source": {
      "name": "VideoPhy2 team",
      "url": "https://videophy2.github.io/",
      "methodologyUrl": "https://github.com/Hritikbansal/videophy/tree/main/VIDEOPHY2"
    },
    "measure": "Joint semantic and physical adherence, %; higher is better.",
    "comparisonRule": "Human-evaluated All subset only. Hard, physical-activity, and object-interaction scores remain separate details.",
    "limitations": [
      "A 2025 research cohort, not a current ranking of video generators.",
      "Physical plausibility is different from cinematic appeal.",
      "Automatic VideoPhy2-eval scores must not be mixed with these human judgments."
    ],
    "coverage": "charted",
    "tags": [
      "video generation",
      "physics",
      "motion",
      "Sora",
      "Wan",
      "Cosmos",
      "world models"
    ]
  },
  "dataset": {
    "benchmarkId": "videophy-2",
    "version": "2 · human evaluation",
    "score": {
      "label": "Prompt + physical adherence",
      "unit": "%",
      "direction": "higher",
      "minimum": 0,
      "maximum": 100
    },
    "source": {
      "name": "VideoPhy2 team",
      "url": "https://videophy2.github.io/",
      "retrievedAt": "2026-09-17T15:07:07.972Z",
      "revision": "sha256:bad716a62a9247323932ca8ea305feb4a2f373ebe01233beece8e3dbea4fbd34"
    },
    "evidenceLabel": "2025 research cohort · human-evaluated",
    "configurationLabel": "Video model · research-paper configuration",
    "comparabilityNote": "Percentage satisfying both semantic adherence and physical commonsense in the All subset. These 2025 human evaluations are a historical diagnostic, not a current video-model buying guide.",
    "points": [
      {
        "id": "wan2-1-14b",
        "label": "Wan2.1-14B",
        "model": "Wan2.1-14B",
        "provider": "Alibaba",
        "harness": null,
        "effort": null,
        "score": 32.6,
        "costUsd": null,
        "uncertainty": null,
        "sourceUrl": "https://videophy2.github.io/",
        "details": [
          {
            "label": "Hard subset · %",
            "value": "21.9"
          },
          {
            "label": "Physical activities · %",
            "value": "31.5"
          },
          {
            "label": "Object interactions · %",
            "value": "36.2"
          },
          {
            "label": "Evaluation",
            "value": "Human judgments"
          }
        ]
      },
      {
        "id": "cogvideox-5b",
        "label": "CogVideoX-5B",
        "model": "CogVideoX-5B",
        "provider": "Z.ai",
        "harness": null,
        "effort": null,
        "score": 25,
        "costUsd": null,
        "uncertainty": null,
        "sourceUrl": "https://videophy2.github.io/",
        "details": [
          {
            "label": "Hard subset · %",
            "value": "0"
          },
          {
            "label": "Physical activities · %",
            "value": "24.6"
          },
          {
            "label": "Object interactions · %",
            "value": "26.1"
          },
          {
            "label": "Evaluation",
            "value": "Human judgments"
          }
        ]
      },
      {
        "id": "cosmos-diff-7b",
        "label": "Cosmos-Diff-7B",
        "model": "Cosmos-Diff-7B",
        "provider": "NVIDIA",
        "harness": null,
        "effort": null,
        "score": 24.1,
        "costUsd": null,
        "uncertainty": null,
        "sourceUrl": "https://videophy2.github.io/",
        "details": [
          {
            "label": "Hard subset · %",
            "value": "10.9"
          },
          {
            "label": "Physical activities · %",
            "value": "22.6"
          },
          {
            "label": "Object interactions · %",
            "value": "27.4"
          },
          {
            "label": "Evaluation",
            "value": "Human judgments"
          }
        ]
      },
      {
        "id": "hunyuan-13b",
        "label": "Hunyuan-13B",
        "model": "Hunyuan-13B",
        "provider": "Tencent",
        "harness": null,
        "effort": null,
        "score": 17.2,
        "costUsd": null,
        "uncertainty": null,
        "sourceUrl": "https://videophy2.github.io/",
        "details": [
          {
            "label": "Hard subset · %",
            "value": "6.2"
          },
          {
            "label": "Physical activities · %",
            "value": "17.6"
          },
          {
            "label": "Object interactions · %",
            "value": "15.9"
          },
          {
            "label": "Evaluation",
            "value": "Human judgments"
          }
        ]
      },
      {
        "id": "videocrafter-2",
        "label": "VideoCrafter-2",
        "model": "VideoCrafter-2",
        "provider": "VideoCrafter-2 project",
        "harness": null,
        "effort": null,
        "score": 10.5,
        "costUsd": null,
        "uncertainty": null,
        "sourceUrl": "https://videophy2.github.io/",
        "details": [
          {
            "label": "Hard subset · %",
            "value": "2.9"
          },
          {
            "label": "Physical activities · %",
            "value": "10.1"
          },
          {
            "label": "Object interactions · %",
            "value": "13.1"
          },
          {
            "label": "Evaluation",
            "value": "Human judgments"
          }
        ]
      },
      {
        "id": "ray2",
        "label": "Ray2",
        "model": "Ray2",
        "provider": "Luma",
        "harness": null,
        "effort": null,
        "score": 20.3,
        "costUsd": null,
        "uncertainty": null,
        "sourceUrl": "https://videophy2.github.io/",
        "details": [
          {
            "label": "Hard subset · %",
            "value": "8.3"
          },
          {
            "label": "Physical activities · %",
            "value": "21"
          },
          {
            "label": "Object interactions · %",
            "value": "18.5"
          },
          {
            "label": "Evaluation",
            "value": "Human judgments"
          }
        ]
      },
      {
        "id": "sora",
        "label": "Sora",
        "model": "Sora",
        "provider": "OpenAI",
        "harness": null,
        "effort": null,
        "score": 23.3,
        "costUsd": null,
        "uncertainty": null,
        "sourceUrl": "https://videophy2.github.io/",
        "details": [
          {
            "label": "Hard subset · %",
            "value": "5.3"
          },
          {
            "label": "Physical activities · %",
            "value": "22.2"
          },
          {
            "label": "Object interactions · %",
            "value": "26.7"
          },
          {
            "label": "Evaluation",
            "value": "Human judgments"
          }
        ]
      }
    ]
  }
}