Qwen2-Audio-7B: benchmark results

Provider: Alibaba. Access: Open.

Aggregate Rating 1135 ± 37, rank #2173 of 2257 rated models, from 16 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
FinBen (Financial LLM)16.01Average Score73.7
HalluAudio - Environmental Sound67.36Average classification accuracy (%) over the seven environme50
Kimi-ASR Internal Testset - Subset 12.31WER (%), lower is better, Moonshot's internal ASR test set s50
Kimi-ASR Internal Testset - Subset 23.24WER (%), lower is better, Moonshot's internal ASR test set s50
TraceAV-Bench (Audio Only)30.74Accuracy (%), macro-average of the 12 general sub-tasks, fou50
FinBen - FinNum0Normalized Score30
HalluAudio - Speech28.72Average classification accuracy (%) over the ten speech task30
MERA Multi - ruSLUn0.1Score (%)22.7
FinBen - MultiFin0Normalized Score15.8
FinBen - QA0Normalized Score15.8
FinBen - FNS0Normalized Score15
MERA Multi - ruEnvAQA14.3Score (%)9.1

Interactive version: theaggregate.ai/model?slug=qwen2-audio-7b · How It Works · Data refreshed daily, snapshot 2026-11-17.