UniVerseBench (Thinking Mode): leaderboard
UniVerseBench (2026) asks large audio-language models expert-verified multiple-choice questions about folk and traditional music recordings in more than 38 languages. Strict accuracy in percent, higher is better: the model answers in thinking mode and needs both a valid reasoning chain and the correct final answer.
Metric: Strict accuracy (%; multiple-choice questions on folk and traditional music recordings from 372 recordings in over 38 languages, 5,042 questions in total, verified by music experts; thinking mode, a valid reasoning chain and the correct final answer both required). Source: arxiv.org. Saturation forecast: Around February 2027. 2 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen3 Omni 30B A3B Instruct | 47.5 | #902 |
| 2 | Qwen2.5-Omni-7B | 33.9 | #1797 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=universebench-thinking-mode · How It Works · Data refreshed daily, snapshot 2026-11-12.