Desktop Task Success on WindowsAgentArena
44.2Task Success Rate
RL + corrected reward estimator (rˆt)
Updated 1mo ago
Evaluation Results
| Method | Links | |
|---|---|---|
| 44.2 | ||
| 40.3 | ||
| 38.3 | ||
| 35.7 | ||
2026.06 | 33.1 |
RL + corrected reward estimator (rˆt)
| Method | Links | |
|---|---|---|
| 44.2 | ||
| 40.3 | ||
| 38.3 | ||
| 35.7 | ||
2026.06 | 33.1 |