GUI Action Critiquing on BBBench (test)
78.25NDCG@8GPT-5
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| GPT-5Model Category=General VLMs2026.05 | 78.25 | 80.49 | 84.52 | 77.38 | 59.81 | 74.96 | |
| BBCritic-7BModel Category=Specialized GUI Policy Models2026.05 | 75.09 | 78.67 | 83.19 | 78.72 | 52.67 | 74.1 | |
| GPT-4oModel Category=General VLMs2026.05 | 74.85 | 76.04 | 80.68 | 80.26 | 47.96 | 69.56 | |
| Gemini 3 ProModel Category=General VLMs2026.05 | 70.92 | 71.86 | 72.62 | 80.66 | 62.47 | 71.35 | |
| BBCritic-3BModel Category=Specialized GUI Policy Models2026.05 | 70.48 | 75.07 | 80.56 | 80.99 | 51.2 | 66.1 | |
| Qwen2.5-VL-7BModel Category=General VLMs2026.05 | 67.04 | 71.53 | 77.78 | 74.32 | 46.9 | 57.62 | |
| Claude Sonnet 4.0Model Category=General VLMs2026.05 | 64.52 | 69.86 | 76.75 | 86.9 | 47.17 | 66.13 | |
| Qwen3-VL-8BModel Category=General VLMs2026.05 | 63 | 69.65 | 75.23 | 75.23 | 52.31 | 74.43 | |
| Gemini 3 FlashModel Category=General VLMs2026.05 | 55.13 | 55.32 | 55.5 | 66.6 | 50.69 | 66.28 | |
| GAIAModel Category=Specialized GUI Policy Models2026.05 | 46.2 | 55.31 | 63.48 | 77.17 | 45.65 | 55.32 | |
| MAI-UIModel Category=Specialized GUI Policy Models2026.05 | 43.11 | 51.75 | 61.83 | 66.62 | 40.23 | 61.4 | |
| UITARS-1.5-7BModel Category=Specialized GUI Policy Models2026.05 | 38.35 | 47.7 | 56.78 | 71.36 | 51.98 | 57.19 | |
| UI-VenusModel Category=Specialized GUI Policy Models2026.05 | 36.12 | 45.55 | 57.03 | 60.39 | 46.8 | 56.89 | |
| GUI-Critic-R1Model Category=Specialized GUI Policy Models2026.05 | 6.4 | 17.62 | 38.51 | 26.53 | 47.56 | 37.03 |