AudioBench - AudioCaps Captioning: leaderboard
AudioBench AudioCaps Captioning slice measuring audio-caption generation quality from sound or speech inputs.
Metric: Mean Score (%). Source: github.com. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | SeaLLMs-Audio-7B | 41.82 | #1018 |
| 2 | Qwen-Audio-Chat | 37.3 | #1244 |
| 3 | Qwen2-Audio-7B-Instruct | 34.07 | #880 |
| 4 | SALMONN-7B | 29.22 | #1293 |
| 5 | Phi-4 Multimodal Instruct | 21.98 | #906 |
Interactive version: theaggregate.ai/benchmark?slug=audiobench-audiocaps-captioning · How It Works · Data refreshed daily, snapshot 2026-10-20.