MMAU-Pro - Music: 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 1,618 music 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
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Audio Flamingo 3 | 61.7 | |
| 2 | Qwen2.5-Omni-7B | 61.5 | #1883 |
| 3 | Qwen2.5-Omni-3B | 60.3 | #3596 |
| 4 | Ming-Lite-Omni-1.5 | 56.2 | #2348 |
| 5 | Audio Flamingo 2 | 55.7 | |
| 6 | Audio Reasoner | 50.1 | #3028 |
| 7 | DeSTA2.5-Audio | 48.2 | #2076 |
| 8 | Phi-4 Multimodal Instruct | 47.8 | #4209 |
| 9 | SALMONN-13B | 47.2 | |
| 10 | Gemma 3n E4B (IT) | 46.4 | #2432 |
| 11 | SALMONN-7B | 44.9 | |
| 12 | Gemma 3n E2B (IT) | 44.1 | #4283 |
| 13 | DeSTA2-8B-beta | 43.3 | |
| 14 | GAMA | 41.2 | #8292 |
| 15 | Mellow | 32.9 |
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-music · How It Works · Data refreshed daily, snapshot 2026-11-17.