Web Visual Understanding on VisualWebBench
74.9WebQAGemini 2.5 Pro
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Gemini 2.5 Pro2025.08 | 74.9 | 70.8 | 95.1 | 80.3 | 91.8 | 96.8 | 90.3 | 93 | 86.6 | |
| Claude Sonnet 42025.08 | 73.3 | 72.6 | 96.2 | 80.7 | 81.1 | 96.1 | 96.3 | 91.2 | 85.9 | |
| UI-TARs-7B-SFT2025.08 | 71.3 | 78.7 | 97.2 | 82.4 | 91.8 | 91.8 | 85.4 | 89.7 | 86 | |
| Qwen2.5-VL-7B2025.08 | 70.8 | 71.7 | 81.4 | 74.6 | 77.5 | 86.8 | 68 | 77.4 | 76 | |
| Web-CogReasoner2025.08 | 67.2 | 72.6 | 97 | 79 | 96.4 | 96.1 | 88.4 | 93.6 | 86.3 |