{ "api_version": "1", "updated": "2026-10-18", "latest_event": "2026-10-17", "rights": "© Frontier Log, all rights reserved.", "benchmark": { "slug": "osworld-verified", "name": "OSWorld-Verified", "family": "osworld", "unit": "%", "higher_is_better": true, "status": "superseded", "benchmark_url": "https://osworld-v1.xlang.ai", "introduced": "2025-07-28", "creator": "XLANG Lab", "introduced_source": "https://xlang.ai/blog/osworld-verified", "settings": [ { "setting": "default", "name": "", "url": "https://frontierlog.ai/benchmark/osworld-verified", "results": 5 } ] }, "total": 5, "results": [ { "id": "gpt-5-4:osworld-verified:default:1", "model": "gpt-5-4", "model_name": "GPT-5.4", "model_date": "2026-03-05", "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4", "benchmark": "osworld-verified", "benchmark_name": "OSWorld-Verified", "setting": "default", "setting_name": "", "score": 75.0, "unit": "%", "printed": "75.0%", "reported_by": "openai", "label": "GPT‑5.4", "source": "https://openai.com/index/introducing-gpt-5-4", "source_title": "Introducing GPT-5.4", "source_date": "2026-03-05", "where": "headline eval table (HTML, first table), row 'OSWorld-Verified', column 'GPT‑5.4'", "own": true, "primary": true, "parallel": false, "notes": "Repeated in the computer-use/vision detail table (row 'OSWorld-Verified', column 'GPT‑5.4'). The '*' footnote on that row belongs to GPT-5.3-Codex's 74.0%. The fetched page text was truncated, so footnotes without a % sign are not visible in the extract.", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-18" }, { "id": "claude-opus-4-7:osworld-verified:default:1", "model": "claude-opus-4-7", "model_name": "Claude Opus 4.7", "model_date": "2026-04-16", "event": "https://frontierlog.ai/event/2026-04-16-claude-opus-4-7", "benchmark": "osworld-verified", "benchmark_name": "OSWorld-Verified", "setting": "default", "setting_name": "", "score": 78.0, "unit": "%", "printed": "78.0%", "reported_by": "anthropic", "label": "Opus 4.7", "source": "https://www.anthropic.com/news/claude-opus-4-7", "source_title": "Introducing Claude Opus 4.7", "source_date": "2026-04-16", "where": "benchmark table image, row 'Agentic computer use / OSWorld-Verified', column 'Opus 4.7'", "own": true, "primary": true, "parallel": false, "notes": "No harness or effort note for OSWorld in the footnotes. Later report: the Opus 4.8 post's footnote says the OSWorld-Verified method changed and restates Opus 4.7 at 82.3% (its table prints 82.8% for Opus 4.7); not recorded here.", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-18", "source_image": "https://frontierlog.ai/sources/527cd462f89b6de4.png", "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/d434d15757c6abac1122af483617741776d5a114-2600x2638.png" }, { "id": "gpt-5-5:osworld-verified:default:1", "model": "gpt-5-5", "model_name": "GPT-5.5", "model_date": "2026-04-23", "event": "https://frontierlog.ai/event/2026-04-23-gpt-5-5", "benchmark": "osworld-verified", "benchmark_name": "OSWorld-Verified", "setting": "default", "setting_name": "", "score": 78.7, "unit": "%", "printed": "78.7%", "reported_by": "openai", "label": "GPT‑5.5", "source": "https://openai.com/index/introducing-gpt-5-5", "source_title": "Introducing GPT-5.5", "source_date": "2026-04-23", "where": "headline eval table (HTML, first table), row 'OSWorld-Verified', column 'GPT‑5.5'", "own": true, "primary": true, "parallel": false, "notes": "Repeated in the computer-use detail table and in the text ('which measures whether a model can operate real computer environments on its own, it reaches 78.7%'). The fetched page text was truncated, so footnotes without a % sign are not visible in the extract.", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-18" }, { "id": "claude-opus-4-8:osworld-verified:default:1", "model": "claude-opus-4-8", "model_name": "Claude Opus 4.8", "model_date": "2026-05-28", "event": "https://frontierlog.ai/event/2026-05-28-claude-opus-4-8", "benchmark": "osworld-verified", "benchmark_name": "OSWorld-Verified", "setting": "default", "setting_name": "", "score": 83.4, "unit": "%", "printed": "83.4%", "reported_by": "anthropic", "label": "Opus 4.8", "source": "https://www.anthropic.com/news/claude-opus-4-8", "source_title": "Introducing Claude Opus 4.8", "source_date": "2026-05-28", "where": "benchmark table image, row 'Agentic computer use / OSWorld-Verified', column 'Opus 4.8'", "own": true, "primary": true, "parallel": false, "notes": "Footnote: Anthropic changed how it runs OSWorld-Verified (details in the system card); scores may not be comparable with earlier Anthropic OSWorld-Verified numbers.", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-18", "source_image": "https://frontierlog.ai/sources/7ad9b784f2579c8a.png", "source_image_original": "https://www-cdn.anthropic.com/images/4zrzovbb/website/a9007019094f217e98cb8261a2765d7646c01708-2600x1392.png" }, { "id": "qwen3-8-max:osworld-verified:default:1", "model": "qwen3-8-max", "model_name": "Qwen3.8-Max", "model_date": "2026-08-03", "event": "https://frontierlog.ai/event/2026-08-03-qwen3-8-max", "benchmark": "osworld-verified", "benchmark_name": "OSWorld-Verified", "setting": "default", "setting_name": "", "score": 86.1, "unit": "%", "printed": "86.1", "reported_by": "alibaba", "label": "Qwen3.8-Max", "source": "https://qwen.ai/blog?id=qwen3.8", "source_title": "Qwen3.8-Max: A New Bar for Coding and Cowork", "source_date": "2026-08-03", "where": "'Full Benchmark Table' (second, multimodal HTML table), section 'Visual Agent & Coding', row 'OSWorld-Verified', column 'Qwen3.8-Max'", "own": true, "primary": true, "parallel": false, "notes": "No harness footnote for this row.", "caveat": "", "human_verified": true, "human_verified_on": "2026-10-18" } ] }