Which AI model scores highest on OSWorld 2.1?

Claude Opus 5.5 by Anthropic holds the best OSWorld 2.1 (full set) result among tracked models, at 81.8% (released Sep 22 2026). Higher scores are better on this benchmark.

What are the top 5 models on OSWorld 2.1?

1. Claude Opus 5.5 (Anthropic) — 81.8%; 2. Claude Fable 5.1 (Anthropic) — 80.7%; 3. Claude Sonnet 5.5 (Anthropic) — 80.1%; 4. Claude Opus 5 (Anthropic) — 74%; 5. Claude Sonnet 5 (Anthropic) — 57%.

How many models have a published OSWorld 2.1 score?

5 tracked models have a published OSWorld 2.1 (full set) score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.

Agentic computer use

OSWorld 2.1full set

An agent operates a desktop across 108 tasks. Each completed checkpoint earns part of the task’s credit, so a model can score even when it only finishes part of a task.

full set offline

Rankings

Higher is better

OSWorld 2.1 — frequently asked questions

: An agent operates a desktop across 108 tasks. Each completed checkpoint earns part of the task’s credit, so a model can score even when it only finishes part of a task.
: Claude Opus 5.5 by Anthropic holds the best OSWorld 2.1 (full set) result among tracked models, at 81.8% (released Sep 22 2026). Higher scores are better on this benchmark.
: 1. Claude Opus 5.5 (Anthropic) — 81.8%; 2. Claude Fable 5.1 (Anthropic) — 80.7%; 3. Claude Sonnet 5.5 (Anthropic) — 80.1%; 4. Claude Opus 5 (Anthropic) — 74%; 5. Claude Sonnet 5 (Anthropic) — 57%.
: 5 tracked models have a published OSWorld 2.1 (full set) score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.