MusicBench-Global: leaderboard
Music understanding from audio (MusicBench, You et al., 2026, introduced with the GaMMA model; unrelated to the Mustango MusicBench dataset): 2,741 multiple-choice questions written by music experts on genre, mood, instruments, singer gender and timbre, artist, language, key, tempo, melody, rhythm, scenario reasoning and content association. Accuracy in percent, higher is better.
Metric: Accuracy (%; 2,741 expert-written multiple-choice questions on global and fine-grained music attributes, scenario reasoning and content association, answered from the audio). Source: arxiv.org. Saturation forecast: Around February 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 76.7 | #262 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=musicbench-global · How It Works · Data refreshed daily, snapshot 2026-11-12.