GUI Agent Navigation on GUI-Odyssey (test)
66Type AccuracyWebFactory-3B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| WebFactory-3BSetting=RL Fine-Tuning2026.03 | 66 | 48.1 | 40.9 | |
| GUI-R1-3BSetting=RL Fine-Tuning2026.03 | 54.8 | 41.5 | 41.3 | |
| QwenVL2.5-3BSetting=Zero-Shot2026.03 | 38.4 | 27.2 | 27.2 | |
| GPT-4oSetting=Zero-Shot2026.03 | 37.5 | 14.2 | 5.4 |