General Multimodal Perception and Reasoning on MMStar
76.6CP ScoreGPT4V
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| GPT4VModel=GPT4V2025.10 | 76.6 | 51.4 | 66.6 | 55.8 | 42.6 | 49.8 | 57.1 | |
| Qwen2.5-VL 7B + GRPOModel=Qwen2.5-VL 7B, Training Strategy=+ GRPO2025.10 | 73.7 | 57.2 | 70.5 | 64.2 | 44.1 | 65.7 | 62.6 | |
| Qwen2.5-VL 7BModel=Qwen2.5-VL 7B, Training Strategy=Base2025.10 | 73.3 | 60.1 | 72.2 | 62 | 44.6 | 67.3 | 63.2 | |
| Qwen2.5-VL 7B + NoisyGRPOModel=Qwen2.5-VL 7B, Training Strategy=+ NoisyGRPO2025.10 | 72.8 | 63.7 | 71.7 | 66.9 | 48.3 | 71.4 | 65.8 | |
| Qwen2.5-VL 7B + SFTModel=Qwen2.5-VL 7B, Training Strategy=+ SFT2025.10 | 71.8 | 58.8 | 70.2 | 60.8 | 45.2 | 67.4 | 62.4 | |
| InternLM-XComposer2Model=InternLM-XComposer22025.10 | 70.8 | 48.8 | 65.2 | 56.4 | 42 | 49.2 | 55.4 | |
| Qwen2.5-VL 3BModel=Qwen2.5-VL 3B, Training Strategy=Base2025.10 | 70.3 | 49.1 | 59.6 | 55.1 | 38.5 | 59.7 | 55.4 | |
| Qwen2.5-VL 3B + SFTModel=Qwen2.5-VL 3B, Training Strategy=+ SFT2025.10 | 70.3 | 49.6 | 59 | 55.4 | 39.2 | 60.4 | 55.7 | |
| Qwen2.5-VL 3B + NoisyGRPOModel=Qwen2.5-VL 3B, Training Strategy=+ NoisyGRPO2025.10 | 69.5 | 53.2 | 66.6 | 60.7 | 37.1 | 62.3 | 58.2 | |
| Qwen2.5-VL 3B + GRPOModel=Qwen2.5-VL 3B, Training Strategy=+ GRPO2025.10 | 68 | 43.3 | 60.4 | 57 | 40 | 58 | 54.5 | |
| CogVLM-ChatModel=CogVLM-Chat2025.10 | 66.8 | 36.8 | 49.2 | 31.2 | 23.6 | 11.6 | 36.5 | |
| LLaVA-Next 34BModel=LLaVA-Next 34B2025.10 | 66.4 | 52 | 62.4 | 46 | 32.4 | 53.6 | 52.1 | |
| Qwen-VL-ChatModel=Qwen-VL-Chat2025.10 | 59.6 | 32 | 50.8 | 29.2 | 22 | 31.6 | 37.5 | |
| LLaVA-1.5 7BModel=LLaVA-1.5 7B2025.10 | 58.8 | 24 | 38.8 | 24 | 13.6 | 22.8 | 30.3 |