Qwen-Audio-Chat: benchmark results

Provider: Alibaba. Access: Open.

Unified ELO 1324 ± 23, rank #2375 of 2656 rated models, from 70 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AudioBench - VoxCeleb Accent48.05Mean Score (%)100
AudioBench - AudioCaps Captioning37.3Mean Score (%)87.5
AudioBench - Clotho AQA61.93Mean Score (%)87.5
AudioBench - IMDA AR Dialogue66.67Mean Score (%)87.5
AudioBench - MELD Emotion50.73Mean Score (%)87.5
AudioBench - AudioCaps QA50.22Mean Score (%)75
AudioBench - MuChoMusic59.06Score (%)75
AudioBench - WavCaps Captioning28.24Mean Score (%)75
AudioBench - LibriSpeech Clean WER2.03WER (%) (lower is better)66.7
AudioBench - TED-LIUM 3 WER4.05WER (%) (lower is better)66.7
AudioBench - IMDA AR Sentence3.93Mean Score (%)62.5
AudioBench - IMDA GR Sentence57.55Mean Score (%)62.5

Interactive version: theaggregate.ai/model?slug=qwen-audio-chat · How It Works · Data refreshed daily, snapshot 2026-09-19.