MMBench
AI benchmark results and model performance
Type: percentageMultimodal
Recorded results
14
Average score
83.49%
Score range
69.20% - 88%
Leading model
88% - Qwen 2.5 Math 72B
Scores over time
Individual benchmark scores plotted by date.
Model results
| Organisation | Models | Reported | Top score | Info | Self reported | Source |
|---|---|---|---|---|---|---|
| Qwen 2.5 Math 72B | - | 88% | inferred high-confidence family alias from qwen2.5-vl-72b (score=0.4700; benches=30) | Yes | Source | |
| Qwen 2.5 Math PRM 72B | - | 88% | inferred family alias from qwen2.5-vl-72b (score=0.4050; benches=30) | Yes | Source | |
| Qwen 2.5 Math RM 72B | - | 88% | inferred family alias from qwen2.5-vl-72b (score=0.4092; benches=30) | Yes | Source | |
| Qwen 2.5 VL 72B Instruct | 26 Jan 2025 | 88% | - | Yes | Source | |
| Phi 4 multimodal instruct | 01 Feb 2025 | 86.70% | - | Yes | Source | |
| Qwen 2.5 0.5B | - | 84.30% | inferred family alias from qwen2.5-vl-7b (score=0.3131; benches=32) | Yes | Source | |
| Qwen 2.5 1.5B | - | 84.30% | inferred family alias from qwen2.5-vl-7b (score=0.3131; benches=32) | Yes | Source | |
| Qwen 2.5 3B | - | 84.30% | inferred family alias from qwen2.5-vl-7b (score=0.3531; benches=32) | Yes | Source | |
| Qwen 2.5 VL 7B Instruct | - | 84.30% | - | Yes | Source | |
| Phi 3 Vision 128K Instruct | - | 81.90% | inferred family alias from phi-3.5-vision-instruct (score=0.3495; benches=9) | Yes | Source | |
| Phi 3.5 vision instruct | 23 Aug 2024 | 81.90% | - | Yes | Source | |
| DeepSeek VL2 Small | 13 Dec 2024 | 80.30% | - | Yes | Source | |
| DeepSeek VL2 | 13 Dec 2024 | 79.60% | - | Yes | Source | |
| DeepSeek VL2 Tiny | 13 Dec 2024 | 69.20% | - | Yes | Source |
