AudioBench Music: leaderboard
Music-focused AudioBench slice covering audio-language model performance on MMAU music questions and MuChoMusic music understanding tasks.
Metric: Average Score (%). Source: huggingface.co. Status: saturated. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen2-Audio-7B-Instruct | 65.89 |
| 2 | Phi-4 Multimodal Instruct | 61.64 |
| 3 | Gemini 1.5 Flash | 58.68 |
| 4 | SALMONN-7B | 53.44 |
| 5 | GPT-4o Audio | 29.04 |
Interactive version: theaggregate.ai/benchmark?slug=audiobench-music · How It Works · Data refreshed daily, snapshot 2026-09-05.