Step Execution on GUI-Odyssey
89.1Step Success RateUI-TARS-7B w/ GPT-4.1
Evaluation Results
| Method | Links | |
|---|---|---|
| UI-TARS-7B w/ GPT-4.1Params=7B w/ -2026.03 | 89.1 | |
| UI-TARS-32BParams=32B2026.03 | 88.6 | |
| UI-TARS-7B w/ TraceR1Params=7B w/ 8B2026.03 | 88.2 | |
| UI-TARS-7BParams=7B2026.03 | 87 | |
| UI-TARS-2BParams=2B2026.03 | 83.4 | |
| OmniParser-v2.0 w/ GPT-4oParams=-2026.03 | 62 | |
| OS-Atlas-7BParams=7B2026.03 | 62 | |
| OS-Atlas-4BParams=4B2026.03 | 56.4 | |
| QwenVL2.5-7BParams=7B2026.03 | 54.5 | |
| QwenVL2.5-3BParams=3B2026.03 | 50.9 | |
| GUI-R1-3BParams=3B2026.03 | 41.3 | |
| GUI-R1-7BParams=7B2026.03 | 38.8 | |
| GPT-4oParams=-2026.03 | 5.4 | |
| Claude-computer-useParams=-2026.03 | 3.1 |