Skip to content

Benchmark

MMAU-Pro

Scores are percentages. Higher is better.

Models measured
12
Scores recorded
12
Sources
2
IQ weight
1.37

Score distribution

59.5%
Mean
62.9%
Median
11.6%
Std dev
39.7%
10th percentile
70.4%
90th percentile
35.4%
Min
35.4%72.2%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.37 for chat models, measured over 10 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Gemini 3.1 Pro Preview· MMAU-Pro data72.2%
2Gemini 3.7 Flash· MMAU-Pro data70.4%
3Gemini 3.6 Flash· MMAU-Pro data69.6%
4MiMo V2 OmniXiaomi MiMo-V2-Omni release table · audio understanding and reasoning· Reported by Xiaomi MiMo· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.69.4%
5Gemini 3.5 Flash· MMAU-Pro data68.5%
6qwen3-omni-30b-a3b-instruct· MMAU-Pro data62.9%
7Gemini 2.5 Flash· MMAU-Pro data59.2%
8GPT Audio· MMAU-Pro data58.8%
9Gemini 2.0 Flash· MMAU-Pro data55.7%
10Qwen2.5-Omni 7B· MMAU-Pro data52.2%
11Gemma 3n E4B IT· MMAU-Pro data39.7%
12Gemma 3n E2b It· MMAU-Pro data35.4%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.