Operating System Task Automation on MacOSArena
0Single App ScoreAguvis-72B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Aguvis-72BStep=502026.01 | 0 | 0 | 0 | |
| GPT-4oStep=502026.01 | 3.57 | 0 | 1.59 | |
| Qwen2.5-VL-72BStep=502026.01 | 7.14 | 0 | 3.17 | |
| Claude-3.7-SonnetStep=502026.01 | 14.29 | 2.86 | 7.94 | |
| UI-TARS-1.5-7BStep=502026.01 | 14.29 | 2.86 | 7.94 | |
| UI-TARS-72B-DPOStep=502026.01 | 14.29 | 5.71 | 9.52 | |
| Qwen3-VL-32B-Inst.Step=502026.01 | 17.86 | 0 | 7.94 | |
| OS-SYMPHONYbackbone=Qwen3-VL-32B-Inst., Step=502026.01 | 32.14 | 8.57 | 19.05 | |
| OS-SYMPHONYbackbone=GPT-5-Mini, Step=502026.01 | 57.14 | 37.14 | 46.03 |