OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments.

369 real computer tasks across Windows, macOS, and Ubuntu requiring GUI interaction. Tests agents operating full desktop apps like spreadsheets, image editors, and terminals. Much harder than web-only benchmarks.

Best published scores.

28 results indexed across 1 metric. Shaded row marks current SOTA; ties broken by submission date.

Primary

success-rate · higher is better

5 steps
of state of the art.

Each row below marks a model that broke the previous record on success-rate. Intermediate submissions are kept in the leaderboard above; only SOTA-setting entries are re-listed here.

Higher scores win. Each subsequent entry improved upon the previous best.

1. Apr 11, 2024Claude Computer UseAnthropic14.90
2. Oct 10, 2024Agent S w/ GPT-4o20.58
3. Apr 1, 2025Agent S2 w/ Claude-3.7-Sonnet34.50
4. May 19, 2025JEDI-7B with o3 planner51
5. Oct 2, 2025Agent S3 w/ bBoN63.50

9 papers
tied to this benchmark.

Every paper below corresponds to at least one row in the leaderboard above. Click through for the arXiv preprint and, when available, the reference implementation.

Neighbouring benchmarks.

Other Web & Desktop Agents datasets on Codesota.

Have a score that beats
this table?

Submit a checkpoint and a reproduction script. We will run it, publish the score, and — if it takes the top — annotate the step on the progress chart with your name.

• 01A public checkpoint or API endpoint
• 02A reproduction script with frozen commit + seed
• 03Declared evaluation environment (Python, deps)
• 04One row per metric declared by this dataset
• 05A contact so we can follow up on discrepancies