Step Execution on AndroidControl High
75.3Step Success RateUI-TARS-7B w/ TraceR1
Evaluation Results
| Method | Links | |
|---|---|---|
| UI-TARS-7B w/ TraceR1Params=7B w/ 8B2026.03 | 75.3 | |
| UI-TARS-7B w/ GPT-4.1Params=7B w/ -2026.03 | 74.8 | |
| UI-TARS-32BParams=32B2026.03 | 74.7 | |
| UI-TARS-7BParams=7B2026.03 | 72.5 | |
| InfiGUI-R1-3BParams=3B2026.03 | 71.1 | |
| UI-TARS-2BParams=2B2026.03 | 68.9 | |
| OmniParser-v2.0 w/ GPT-4oParams=-2026.03 | 58.8 | |
| GUI-R1-7BParams=7B2026.03 | 51.7 | |
| QwenVL2.5-7BParams=7B2026.03 | 47.1 | |
| GUI-R1-3BParams=3B2026.03 | 46.5 | |
| QwenVL2.5-3BParams=3B2026.03 | 38.9 | |
| OS-Atlas-7BParams=7B2026.03 | 29.8 | |
| OS-Atlas-4BParams=4B2026.03 | 22.7 | |
| GPT-4oParams=-2026.03 | 21.2 | |
| Claude-computer-useParams=-2026.03 | 12.5 |