GUI Grounding on OSWorld G-Refine v1.0 (test)
76.1Overall Success RateHolo2
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Holo2Model Scale=≥30B2026.06 | 76.1 | — | — | — | — | — | |
| MAI-UI-32BZoom-In=true2025.12 | 75 | 79.3 | 79.7 | 84.2 | 64.4 | — | |
| MAI-UI-32BZoom-In=false2025.12 | 73.9 | 79.7 | 79.4 | 81 | 61.7 | — | |
| MAI-UI-8BZoom-In=true2025.12 | 72.9 | 79.3 | 78.8 | 84.2 | 59.7 | — | |
| VISTAModel Scale=≥30B, Inference Method=Multi-view (+ MVP)2026.06 | 72.5 | — | — | — | — | — | |
| GTA1-32B2025.12 | 72.2 | 63.2 | 83.6 | 84.4 | 70.5 | 0 | |
| GTA1Model Scale=≥30B2026.06 | 72.2 | — | — | — | — | — | |
| VISTAModel Scale=≥30B, Inference Method=Single-view2026.06 | 72 | — | — | — | — | — | |
| VISTAModel Scale=≈8B, Inference Method=Multi-view (+ MVP)2026.06 | 70.9 | — | — | — | — | — | |
| VISTAModel Scale=≈8B, Inference Method=Single-view2026.06 | 70.8 | — | — | — | — | — | |
| OpenCUA-32B2025.12 | 70.2 | 63.2 | 79.9 | 84.9 | 62.1 | 7.4 | |
| OpenCUAModel Scale=≥30B2026.06 | 70.2 | — | — | — | — | — | |
| Holo2Model Scale=≈8B2026.06 | 70.1 | — | — | — | — | — | |
| Step-GUIModel Scale=≈8B2026.06 | 70 | — | — | — | — | — | |
| VISTAModel Scale=≈4B, Inference Method=Multi-view (+ MVP)2026.06 | 69.6 | — | — | — | — | — | |
| UI-VenusModel Scale=72B2026.06 | 69.5 | — | — | — | — | — | |
| VISTAModel Scale=≈4B, Inference Method=Single-view2026.06 | 69.4 | — | — | — | — | — | |
| Qwen3-VLModel Scale=≥30B2026.06 | 69.3 | — | — | — | — | — | |
| Qwen3-VL-32BEvaluated by authors=true2025.12 | 69 | 77.4 | 73.6 | 76.3 | 57.7 | — | |
| MAI-UI-8BZoom-In=false2025.12 | 68.6 | 77.4 | 73 | 78.3 | 55.7 | — | |
| MAI-UIModel Scale=≈8B2026.06 | 68.6 | — | — | — | — | — | |
| Qwen3-VLModel Scale=≈4B2026.06 | 67.9 | — | — | — | — | — | |
| GTA1-7B2025.12 | 67.7 | 63.2 | 82.1 | 74.2 | 70.5 | 0 | |
| GTA1Model Scale=≈8B2026.06 | 67.7 | — | — | — | — | — | |
| Step-GUIModel Scale=≈4B2026.06 | 66.9 | — | — | — | — | — | |
| MAI-UI-2BZoom-In=true2025.12 | 66.3 | 71.3 | 71.8 | 78.3 | 51 | — | |
| Qwen3-VL-8BEvaluated by authors=true2025.12 | 64.4 | 73.9 | 68.2 | 73.1 | 54.4 | — | |
| Qwen3-VLModel Scale=≈8B2026.06 | 64.4 | — | — | — | — | — | |
| UI-TARS-1.5-7B2025.12 | 64.2 | 52.6 | 75.4 | 72.4 | 66.7 | 0 | |
| UI-TARS-1.5Model Scale=≈8B2026.06 | 64.2 | — | — | — | — | — | |
| Jedi-7B2025.12 | 63.8 | — | — | — | — | — | |
| MAI-UI-2BZoom-In=false2025.12 | 63.5 | 70.9 | 69.1 | 72.7 | 47.7 | — | |
| UI-VenusModel Scale=≈8B2026.06 | 61.7 | — | — | — | — | — | |
| Jedi-3B2025.12 | 61 | — | — | — | — | — | |
| Qwen2.5-VL-32B2025.12 | 59.6 | 57.9 | 70.2 | 73.8 | 49.2 | 0 | |
| Operator2025.12 | 57.8 | — | — | — | — | — | |
| Qwen3-VL-2BEvaluated by authors=true2025.12 | 57.4 | 69.3 | 60.9 | 69.2 | 45 | — |