GUI Agent Task Completion on DataScope V1
18.8Success Rate (SR)GPT-5.2+ UIIns
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GPT-5.2+ UIInsParams=–2026.06 | 18.8 | 35.4 | 38.4 | |
| ATMem-UIParams=8B2026.06 | 6.2 | 11.7 | 15.7 | |
| GUI-Owl-1.5Params=8B2026.06 | 3.1 | 4.7 | 7.3 | |
| MAI-UIParams=8B2026.06 | 3.1 | 8.8 | 10.7 | |
| GELab-ZeroParams=4B2026.06 | 0 | 4.2 | 3.9 | |
| UI-TARS-1.5Params=7B2026.06 | 0 | 1.6 | 1.7 | |
| GUI-OwlParams=7B2026.06 | 0 | 3.6 | 5.7 | |
| UI-VenusParams=7B2026.06 | 0 | 1.6 | 3.8 | |
| GUI-OwlParams=32B2026.06 | 0 | 2.6 | 2.2 | |
| UI-Venus-1.5-A3BParams=30B2026.06 | 0 | 1.6 | 4.1 |