Grounding & counting on Visual Web Bench
88Pass@1Seed 1.5-VL
Evaluation Results
| Method | Links | |
|---|---|---|
| Seed 1.5-VLthinking=false, decoding=greedy2025.05 | 88 | |
| Seed 1.5-VLthinking=true, decoding=greedy2025.05 | 87.3 | |
| Gemini 1.5 Prothinking=true, decoding=greedy2025.05 | 87.3 | |
| Claude 3.7 Sonnetthinking=true, decoding=sampling2025.05 | 85.9 | |
| Qwen 2.5-VL 72Bthinking=false, decoding=greedy2025.05 | 82.3 | |
| OpenAI o1thinking=true, decoding=greedy2025.05 | 80.9 | |
| GPT-4othinking=false, decoding=greedy2025.05 | 80.2 |