Multimodal Understanding on MM-Vet v2
71.8MM-Vet v2 ScoreClaude3.5-Sonnet
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude3.5-Sonnet2024.12 | 71.8 | |
| GPT-4o (0806)Model Version=08062024.12 | 71 | |
| VLSI-7BSize=7B2024.12 | 70 | |
| Qwen2-VL-72BSize=72B2024.12 | 68.7 | |
| InternVL2-76BSize=76B2024.12 | 68.4 | |
| Gemini-1.5-Pro2024.12 | 66.9 | |
| GPT-4V (0409)Model Version=04092024.12 | 66.3 | |
| NDBackbone Model=Qwen3-VL2025.08 | 66 | |
| NDModel=Qwen3-VL, Note=no-defense baseline2025.08 | 66 | |
| PRISMBackbone Model=Qwen3-VL2025.08 | 62.5 | |
| PRISMModel=Qwen3-VL2025.08 | 62.5 | |
| VLSI-2BSize=2B2024.12 | 60.8 | |
| Phantom-7BSize=7B2024.12 | 60.6 | |
| NDBackbone Model=Qwen2.5-VL2025.08 | 59 | |
| NDModel=Qwen2.5-VL, Note=no-defense baseline2025.08 | 59 | |
| LLaVA-NeXT-34BSize=34B2024.12 | 50.9 | |
| PRISMBackbone Model=Qwen2.5-VL2025.08 | 49.2 | |
| PRISMModel=Qwen2.5-VL2025.08 | 49.2 | |
| Phantom-1.8BSize=1.8B2024.12 | 46.3 | |
| PRISMBackbone Model=LLaVA-1.62025.08 | 27.9 | |
| PRISMModel=LLaVA-1.62025.08 | 27.9 | |
| NDBackbone Model=LLaVA-1.62025.08 | 20.9 | |
| NDModel=LLaVA-1.6, Note=no-defense baseline2025.08 | 20.9 |