{
  "api_version": "1",
  "updated": "2026-10-26",
  "latest_event": "2026-10-26",
  "rights": "© Frontier Log, all rights reserved.",
  "model": {
    "slug": "muse-wonder",
    "name": "Muse Wonder",
    "lab": "meta",
    "date": "2026-10-12",
    "source": "https://research.meta.ai/blog/muse-wonder",
    "source_title": "Muse Wonder: A Frontier Model For Personal & Professional AI Agents",
    "flagship": false,
    "event": "https://frontierlog.ai/event/2026-10-12-muse-wonder",
    "releases": [],
    "variant_of": null,
    "made_by": "Meta",
    "url": "https://frontierlog.ai/model/muse-wonder",
    "results": [
      {
        "id": "muse-wonder:deepswe-1-1:default:1",
        "model": "muse-wonder",
        "model_name": "Muse Wonder",
        "model_date": "2026-10-12",
        "event": "https://frontierlog.ai/event/2026-10-12-muse-wonder",
        "benchmark": "deepswe-1-1",
        "benchmark_name": "DeepSWE v1.1",
        "setting": "default",
        "setting_name": "",
        "score": 78.8,
        "unit": "%",
        "printed": "78.8",
        "reported_by": "meta",
        "run_by": null,
        "label": "Muse Wonder (max)",
        "source": "https://research.meta.ai/blog/muse-wonder",
        "source_title": "Muse Wonder: A Frontier Model For Personal & Professional AI Agents",
        "source_date": "2026-10-12",
        "where": "evaluation table (image), row 'DeepSWE 1.1', column 'Muse Wonder (max)'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "run by Datacurve; max effort",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-10-26",
        "corroborated_by": [
          {
            "url": "https://research.meta.ai/static/muse-wonder-model-card",
            "name": "Muse Wonder and Muse Wonder Full card",
            "title": "Muse Wonder model card",
            "lab": "meta",
            "date": "2026-10-12",
            "printed": "78.8",
            "where": "Model card capabilities table, row \"DeepSWE 1.1\", column \"Muse Wonder (max)\". Table is an image embedded on page 3 of the model card PDF (PDF dated 2026-10-23; sha256 is of the PDF bytes). Card methodology: max reasoning effort; DeepSWE uses Datacurve third-party results; OSWorld V2.1 = 108 workflows, primary metric mean per-task partial reward, strict binary completion secondary."
          }
        ],
        "source_image": "https://frontierlog.ai/sources/4a9f891f33ba1b87.webp",
        "source_image_original": "https://research.meta.ai/articles/introducing-muse-wonder/images/muse-evaluation-table-light-v6.webp"
      },
      {
        "id": "muse-wonder:osworld-2-1:partial:1",
        "model": "muse-wonder",
        "model_name": "Muse Wonder",
        "model_date": "2026-10-12",
        "event": "https://frontierlog.ai/event/2026-10-12-muse-wonder",
        "benchmark": "osworld-2-1",
        "benchmark_name": "OSWorld 2.1",
        "setting": "partial",
        "setting_name": "Full set, partial score",
        "score": 72.6,
        "unit": "%",
        "printed": "72.6",
        "reported_by": "meta",
        "run_by": null,
        "label": "Muse Wonder (max)",
        "source": "https://research.meta.ai/blog/muse-wonder",
        "source_title": "Muse Wonder: A Frontier Model For Personal & Professional AI Agents",
        "source_date": "2026-10-12",
        "where": "evaluation table (image), row 'OSWorld V2.1', column 'Muse Wonder (max)', the value marked '(partial)'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "max effort; mean per-task partial reward; 108 tasks",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-10-26",
        "corroborated_by": [
          {
            "url": "https://research.meta.ai/static/muse-wonder-model-card",
            "name": "Muse Wonder and Muse Wonder Full card",
            "title": "Muse Wonder model card",
            "lab": "meta",
            "date": "2026-10-12",
            "printed": "72.6",
            "where": "Model card capabilities table, row \"OSWorld V2.1 (partial)\", column \"Muse Wonder (max)\". Table is an image embedded on page 3 of the model card PDF (PDF dated 2026-10-23; sha256 is of the PDF bytes). Card methodology: max reasoning effort; DeepSWE uses Datacurve third-party results; OSWorld V2.1 = 108 workflows, primary metric mean per-task partial reward, strict binary completion secondary."
          }
        ],
        "source_image": "https://frontierlog.ai/sources/4a9f891f33ba1b87.webp",
        "source_image_original": "https://research.meta.ai/articles/introducing-muse-wonder/images/muse-evaluation-table-light-v6.webp"
      },
      {
        "id": "muse-wonder:osworld-2-1:strict:1",
        "model": "muse-wonder",
        "model_name": "Muse Wonder",
        "model_date": "2026-10-12",
        "event": "https://frontierlog.ai/event/2026-10-12-muse-wonder",
        "benchmark": "osworld-2-1",
        "benchmark_name": "OSWorld 2.1",
        "setting": "strict",
        "setting_name": "Full set, strict pass rate",
        "score": 34.3,
        "unit": "%",
        "printed": "34.3",
        "reported_by": "meta",
        "run_by": null,
        "label": "Muse Wonder (max)",
        "source": "https://research.meta.ai/blog/muse-wonder",
        "source_title": "Muse Wonder: A Frontier Model For Personal & Professional AI Agents",
        "source_date": "2026-10-12",
        "where": "evaluation table (image), row 'OSWorld V2.1', column 'Muse Wonder (max)', the value marked '(binary)'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "max effort; strict binary completion; 108 tasks",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-10-26",
        "corroborated_by": [
          {
            "url": "https://research.meta.ai/static/muse-wonder-model-card",
            "name": "Muse Wonder and Muse Wonder Full card",
            "title": "Muse Wonder model card",
            "lab": "meta",
            "date": "2026-10-12",
            "printed": "34.3",
            "where": "Model card capabilities table, row \"OSWorld V2.1 (binary)\", column \"Muse Wonder (max)\". Table is an image embedded on page 3 of the model card PDF (PDF dated 2026-10-23; sha256 is of the PDF bytes). Card methodology: max reasoning effort; DeepSWE uses Datacurve third-party results; OSWorld V2.1 = 108 workflows, primary metric mean per-task partial reward, strict binary completion secondary."
          }
        ],
        "source_image": "https://frontierlog.ai/sources/4a9f891f33ba1b87.webp",
        "source_image_original": "https://research.meta.ai/articles/introducing-muse-wonder/images/muse-evaluation-table-light-v6.webp"
      }
    ]
  }
}