AudioBench - Alpaca Audio: leaderboard
AudioBench Alpaca Audio slice measuring instruction following and response quality for audio-conditioned prompts.
Metric: Mean Score (%). Source: github.com. Saturation forecast: Estimated already saturated. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen2-Audio-7B-Instruct | 57.1 | #880 |
| 2 | Phi-4 Multimodal Instruct | 25.8 | #906 |
| 3 | SALMONN-7B | 17.2 | #1293 |
| 4 | Qwen-Audio-Chat | 9.8 | #1244 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=audiobench-alpaca-audio · How It Works · Data refreshed daily, snapshot 2026-10-20.