Multi-step Reasoning on MMMU (Accuracy)
68.3Accuracy (MMMU Multi-step Reasoning)Claude-3.5
Evaluation results
| Method | Links | |
|---|---|---|
| Claude-3.5Size=-2026.05 | 68.3 | |
| Qwen2.5-VL-InstructSize=72B2026.05 | 67 | |
| VL-RethinkerSize=7B2026.05 | 56.7 | |
| MoCASize=7B2026.05 | 54.8 | |
| Qwen2.5-VL-InstructSize=7B2026.05 | 54.3 | |
| GPT-4oSize=-2026.05 | 51.9 | |
| Pixel ReasonerSize=7B2026.05 | 50.8 | |
| Llava-OVSize=7B2026.05 | 48.8 | |
| DeepEyesSize=7B2026.05 | 45.2 | |
| GPT-4o-miniSize=-2026.05 | 45.1 | |
| mPLUG-Owl3Size=7B2026.05 | 42.9 | |
| DocopilotSize=8B2026.05 | 36.6 | |
| R1-VLSize=7B2026.05 | 7.8 |