GUI Reasoning on OmniAct-Web
87.24Type Success RateCAPO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CAPOPrompting=zero-shot2025.12 | 87.24 | 74.02 | 74.16 | |
| GRPOPrompting=zero-shot2025.12 | 79.02 | 71.1 | 70.76 | |
| QwenVL2.5-3BPrompting=zero-shot2025.12 | 50.63 | 46.89 | 47.02 | |
| Os-Atlas-4BPrompting=zero-shot2025.12 | 46.74 | 49.24 | 22.99 |