Benchmarks
OSWorld 2
Models scored
9
Best score
0.21
Metric
Binary accuracy
OSWorld 2 is an AI evaluation tracked by Epoch AI, with 9 scored model versions on record. The scores here are reported by third parties rather than produced by Epoch AI running the evaluation itself.
Scored models span February 2026 to June 2026. The best score recorded is 0.21, measured as Binary accuracy. Models come from Anthropic, OpenAI, MiniMax, Moonshot.
Full record
- Score metric
- Binary accuracy
- Models scored
- 9
- Best score recorded
- 0.21
- Earliest model scored
- Feb 17, 2026
- Most recent model scored
- Jun 1, 2026
- Third-party reported
- Yes
Leaderboard
| 01 | claude-opus-4-8_max | Anthropic | 0.21 |
| 02 | claude-opus-4-7_max | Anthropic | 0.18 |
| 03 | gpt-5.5_xhigh | OpenAI | 0.13 |
| 04 | claude-sonnet-4-6_medium | Anthropic | 0.09 |
| 05 | claude-sonnet-4-6_max | Anthropic | 0.08 |
| 06 | MiniMax-M3 | MiniMax | 0.05 |
| 07 | kimi-k2.6 | Moonshot | 0.05 |
More benchmarks
SourceEpoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/benchmarks. Licensed under CC BY 4.0.
← All benchmarks