MMAU-Pro - Sound: leaderboard

Audio-language models answer human-written questions about real-world speech, sound and music clips of up to 10 minutes; the score is accuracy on the 1,654 sound questions, from the authors' runs. Unlike the original MMAU, clips are long, can be mixed or multiple, and include spatial audio, voice chat and open-ended questions.

Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 18 models tracked. Updated 17 Nov 2026.

Top models

#ModelScoreOverall rank
1Audio Flamingo 355.9
2R1-AQA47.9#5707
3Ming-Lite-Omni-1.547.9#2348
4Qwen2.5-Omni-7B47.6#1883
5GAMA45.4#8292
6SALMONN-13B43.6
7Gemma 3n E4B (IT)42.4#2432
8Gemma 3n E2B (IT)40.1#4283
9Audio Flamingo 239.5
10Qwen2.5-Omni-3B38.5#3596
11DeSTA2.5-Audio35.7#2076
12Baichuan-Omni-1.534.6#3064
13Audio Reasoner34.2#3028
14SALMONN-7B32.2
15DeSTA2-8B-beta31

No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.

Interactive version: theaggregate.ai/benchmark?slug=mmau-pro-sound · How It Works · Data refreshed daily, snapshot 2026-11-17.