GUI agent task on OSWORLD clean (Overall)
30.3Pass@8 Success RateUI-TARS-1.5-7B + ENVS
Evaluation Results
| Method | Links | |
|---|---|---|
| UI-TARS-1.5-7B + ENVSModel=UI-TARS-1.5-7B + ENVS, Train data=clean trajectories2026.06 | 30.3 | |
| UI-TARS-1.5-7B + ARPOModel=UI-TARS-1.5-7B + ARPO, Train data=clean rollouts2026.06 | 26.7 | |
| UI-TARS-1.5-7BModel=UI-TARS-1.5-7B, Train data=—2026.06 | 22.7 |