GUI Navigation on Mind2Web Cross-Task
66Element AccuracyMindAct
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| MindAct2026.03 | 66 | 36.2 | 60.6 | 2 | |
| Upper BoundBackbone=Qwen2.5VL-3B2026.03 | 59.64 | 55.79 | 70.8 | — | |
| FASTVBackbone=Qwen2.5VL-3B2026.03 | 58.73 | 54.83 | 69.46 | — | |
| SparseVLMBackbone=Qwen2.5VL-3B2026.03 | 58.68 | 54.97 | 69.42 | — | |
| RandomBackbone=Qwen2.5VL-3B2026.03 | 58.58 | 54.78 | 69.15 | — | |
| DivPruneBackbone=Qwen2.5VL-3B2026.03 | 58.44 | 54.78 | 70.48 | — | |
| PDropBackbone=Qwen2.5VL-3B2026.03 | 58.34 | 54.64 | 69.34 | — | |
| DartBackbone=Qwen2.5VL-3B2026.03 | 57.09 | 53.05 | 69.32 | — | |
| HMTBackbone=GPT-42026.03 | 54.2 | 48.5 | 63.5 | 4.6 | |
| AWMSetting=offline2026.03 | 50.6 | 45.1 | 57.3 | 4.8 | |
| AWMSetting=online2026.03 | 50 | 43.6 | 56.4 | 4 | |
| FASTVBackbone=Qwen2VL-2B2026.03 | 49.11 | 46.12 | 64.36 | — | |
| Upper BoundBackbone=Qwen2VL-2B2026.03 | 48.96 | 45.64 | 63.33 | — | |
| RandomBackbone=Qwen2VL-2B2026.03 | 48.86 | 45.79 | 62.4 | — | |
| PDropBackbone=Qwen2VL-2B2026.03 | 48.29 | 45.21 | 62.89 | — | |
| DartBackbone=Qwen2VL-2B2026.03 | 47.66 | 44.68 | 62.7 | — | |
| DivPruneBackbone=Qwen2VL-2B2026.03 | 47.61 | 44.73 | 62.87 | — | |
| SparseVLMBackbone=Qwen2VL-2B2026.03 | 47.57 | 44.3 | 61.58 | — | |
| MiniCPM-GUILanguage Model Size=Larger2024.06 | 36.9 | 33.2 | — | — | |
| SeeClick2024.06 | 28.3 | 25.5 | — | — | |
| SeeClick2024.06 | 28.3 | 25.5 | — | — | |
| Qwen-GUI2024.06 | 27.9 | 24.4 | — | — | |
| Qwen-GUI2024.06 | 27.9 | 24.4 | — | — | |
| MiniCPM-GUILanguage Model Size=Standard2024.06 | 23.8 | 20.8 | — | — | |
| MiniCPM-GUI2024.06 | 23.8 | 20.8 | — | — | |
| Qwen-VL2024.06 | 23.2 | 20.3 | — | — | |
| Fuyu-GUI2024.06 | 19.1 | 15.6 | — | — | |
| Fuyu-GUI2024.06 | 19.1 | 15.6 | — | — | |
| MiniCPM-V2024.06 | 11 | 8.5 | — | — | |
| Fuyu-8BBackbone=Fuyu-8B2024.06 | 8.3 | 6.6 | — | — |