Reasoning
MMMU
Multimodal Multidiscipline Understanding — multimodal reasoning over academic images.
11 models published a score
| # | Model | Company | Score |
|---|---|---|---|
| 1 | Nova Premier | Amazon | 87.4 |
| 2 | Qwen3.6-Plus | Alibaba | 86.0 |
| 3 | Doubao Seed 2.0 Pro | ByteDance | 85.4 |
| 4 | Qwen3.5-397B-A17B | Alibaba | 85.0 |
| 5 | Qwen3.5-Omni-Plus | Alibaba | 82.0 |
| 6 | Step-3 | StepFun | 74.2 |
| 7 | Gemma 4 26B-A4B | Google DeepMind | 73.8 |
| 8 | Llama 4 Maverick | Meta | 73.4 |
| 9 | Llama 4 Scout | Meta | 69.4 |
| 10 | Gemma 4 E4B | Google DeepMind | 52.6 |
| 11 | Gemma 4 E2B | Google DeepMind | 44.2 |