OSWorld osworld Leaderboard
Auto-discovered from papers reporting OSWorld (Success rate). · Metric: Success rate (higher is better) · 🟢 Updated 2h ago
| # | Model | Success rate | Paper |
|---|---|---|---|
| 1 | Learning from Failure: Inference-Time Self-Improvement for Computer-Use Agents | 48.90 | link |
| 2 | ProCUA-SFT Technical Report | 45.00 | link |