OSWorld 2 (Aug 8) Full, 500-step Binary Accuracy
Agents · 2026-08-08
OSWorld 2.0 v2026.08.08 full 108-task release, binary accuracy with 500 steps. Separate from June tasks and the offline subset.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5 | 31.4 |
| GPT-5.6 Sol | 27.3 |