Multimodal Reasoning on MMMU-Pro (test)
55AccuracyClaude-3.5 Sonnet
Evaluation results
| Method | Links | |
|---|---|---|
| Claude-3.5 Sonnet2026.04 | 55 | |
| GPT-4o2026.04 | 54 | |
| Gemini-1.5-Pro2026.04 | 49.4 | |
| InternVL2.5-8B2026.04 | 38.2 | |
| Qwen2.5-VL-7B + Saliency-R1Training Stage=Saliency-R12026.04 | 37.6 | |
| Qwen2.5-VL-7BTraining Stage=Base2026.04 | 36.2 | |
| Qwen2.5-VL-7B + SFTTraining Stage=SFT2026.04 | 35.5 | |
| InternVL2-8B2026.04 | 32.5 | |
| Qwen2.5-VL-3B + Saliency-R1Training Stage=Saliency-R12026.04 | 31.1 | |
| Qwen2.5-VL-3BTraining Stage=Base2026.04 | 30.8 | |
| Qwen2.5-VL-3B + SFTTraining Stage=SFT2026.04 | 29.9 | |
| MiniCPM-V-2.6-8B2026.04 | 27.2 | |
| Insight-V-8B2026.04 | 24.9 | |
| MiniCPM-Llama-V-2.5-8B2026.04 | 19.6 |