GUI reasoning on GUI-Act-Web
87.73Type Success RateCAPO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CAPOPrompting=zero-shot2025.12 | 87.73 | 85.85 | 85.85 | |
| GRPOPrompting=zero-shot2025.12 | 85.1 | 82.36 | 70.23 | |
| Os-Atlas-4BPrompting=zero-shot2025.12 | 79.22 | 58.57 | 42.62 | |
| QwenVL2.5-3BPrompting=zero-shot2025.12 | 56.1 | 64.28 | 55.61 |