Visual Grounding on ScreenSpot V2 (Device/Input Modality Breakdown)
90.5Average AccuracyLPO-8B
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| LPO-8BCategory=GUI-specific (RL)2025.06 | 90.5 | 97.9 | 82.9 | 95.9 | 86.4 | 95.6 | 84.2 | |
| UI-TARS-72BCategory=GUI-specific (SFT)2025.06 | 90.3 | 94.8 | 86.3 | 91.2 | 87.9 | 91.5 | 87.7 | |
| SE-GUI-7BCategory=GUI-specific (RL)2025.06 | 90.3 | — | — | — | — | — | — | |
| Qwen2.5-VL-7BCategory=General open-source2025.06 | 88.8 | 97.6 | 87.2 | 90.2 | 74.2 | 93.2 | 81.3 | |
| UGround-V1-7BCategory=GUI-specific (SFT)2025.06 | 87.6 | 95 | 83.3 | 95 | 77.8 | 92.1 | 77.2 | |
| Aguvis-7BCategory=GUI-specific (SFT)2025.06 | 86 | 95.5 | 77.3 | 95.4 | 77.9 | 91 | 72.4 | |
| UI-TARS-2BCategory=GUI-specific (SFT)2025.06 | 84.7 | 95.2 | 79.1 | 90.7 | 68.6 | 87.2 | 78.3 | |
| OS-Atlas-7BCategory=GUI-specific (SFT)2025.06 | 84.1 | 95.2 | 75.8 | 90.7 | 63.6 | 90.6 | 77.3 | |
| Qwen2.5-VL-3BCategory=General open-source2025.06 | 80.9 | 93.4 | 73.5 | 88.1 | 58.6 | 88 | 71.4 | |
| GPT-4o + OmniParser-v2Category=Proprietary2025.06 | 80.7 | 95.5 | 74.6 | 92.3 | 60.9 | 88 | 59.6 | |
| OS-Atlas-4BCategory=GUI-specific (SFT)2025.06 | 71.9 | 87.2 | 59.7 | 72.7 | 46.4 | 85.9 | 63.1 | |
| OperatorCategory=Proprietary2025.06 | 70.5 | 47.3 | 41.5 | 90.2 | 80.3 | 92.8 | 84.3 | |
| Magma-8BCategory=GUI-specific (SFT)2025.06 | 61.5 | 62.8 | 53.4 | 80 | 57.9 | 67.5 | 47.3 | |
| SeeClick-9.6BCategory=GUI-specific (SFT)2025.06 | 55.1 | 78.4 | 50.7 | 70.1 | 29.3 | 55.2 | 32.5 |