Grounding on OSWorld-G-Refine (test)
76.4Overall Success RateUI-Venus-1.5-30B-A3B
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| UI-Venus-1.5-30B-A3BVersion=1.5, Model Size=30B-A3B2026.06 | 76.4 | — | — | — | — | — | |
| UI-Venus-1.5-8BVersion=1.5, Model Size=8B2026.06 | 74.1 | — | — | — | — | — | |
| MAI-UI-32BModel Size=32B2026.06 | 73.9 | — | — | — | — | — | |
| Step-GUI-8BModel Size=8B2026.06 | 70 | — | — | — | — | — | |
| GUI-Owl-1.5-32B-InstructVersion=1.5, Model Size=32B, Training=Instruct2026.06 | 69.7 | 68.5 | 68.5 | 71 | 70.1 | 68.5 | |
| UI-Venus-Ground-72BType=Ground, Model Size=72B2026.06 | 69.5 | — | — | — | — | — | |
| Qwen3-VL-30B-A3BModel Size=30B-A3B2026.06 | 69.3 | — | — | — | — | — | |
| GUI-Owl-1.5-8B-InstructVersion=1.5, Model Size=8B, Training=Instruct2026.06 | 69.3 | 73.1 | 69.4 | 71 | 74.8 | 42.5 | |
| MAI-UI-8BModel Size=8B2026.06 | 68.6 | — | — | — | — | — | |
| GUI-Owl-1.5-4B-InstructVersion=1.5, Model Size=4B, Training=Instruct2026.06 | 68.4 | 73.1 | 71.8 | 72.9 | 74.1 | 29.6 | |
| Qwen3-VL-8BModel Size=8B2026.06 | 67 | — | — | — | — | — | |
| Step-GUI-4BModel Size=4B2026.06 | 66.9 | — | — | — | — | — | |
| GUI-Owl-1.5-32B-ThinkingVersion=1.5, Model Size=32B, Training=Thinking2026.06 | 64.7 | 68.9 | 69.7 | 72.6 | 72.1 | 5.6 | |
| UI-TARS-1.5-7BVersion=1.5, Model Size=7B2026.06 | 64.2 | 52.6 | 75.4 | 72.4 | 66.7 | 0 | |
| Xiaomi-GUI-0-30B-A3BModel Size=30B-A3B2026.06 | 64.2 | 52.6 | 72.4 | 81.3 | 54.5 | 0 | |
| UI-Venus-Ground-7BType=Ground, Model Size=7B2026.06 | 61.7 | — | — | — | — | — | |
| GUI-Owl-1.5-8B-ThinkingVersion=1.5, Model Size=8B, Training=Thinking2026.06 | 59.8 | 64.7 | 61.1 | 64.9 | 65.3 | 20.4 |