Visual Understanding on VStar
85.86AccuracyQwen2.5-VL
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-VLSize=72B2025.10 | 85.86 | |
| InternVL3.5Size=38B2025.10 | 76.96 | |
| Qwen2.5-VLSize=7B2025.10 | 76.44 | |
| GPT-5Size=-2025.10 | 71.73 | |
| Gemini 2.5 FlashSize=-2025.10 | 70.68 | |
| InternVL3.5Size=8B2025.10 | 68.59 | |
| ThinkMorphSize=7B, think mode=true2025.10 | 67.02 | |
| GPT-4oSize=-2025.10 | 61.78 | |
| BagelSize=7B, think mode=true2025.10 | 55.49 | |
| Janus-proSize=7B2025.10 | 38.22 | |
| ChameleonSize=7B2025.10 | 28.27 |