MOSS-Audio-8B-Instruct: benchmark results
Provider: Other.
Aggregate Rating 1524 ± 46, rank #2024 of 10916 rated models, from 9 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| BASIS-Bench - ASR Fair-Speech Accent | 90.8 | 100-WER | 57.1 |
| BASIS-Bench - ASR Fair-Speech Gender | 89.4 | 100-WER | 57.1 |
| BASIS-Bench - ASR SD-QA Accent | 92.6 | 100-WER | 57.1 |
| BASIS-Bench - SQA SD-QA Accent | 75.5 | Accuracy (%) | 42.9 |
| BASIS-Bench - ASR Spoken StereoSet Age | 99.1 | 100-WER | 35.7 |
| BASIS-Bench - SQA LibriSQA Gender | 75.5 | Accuracy (%) | 28.6 |
| VoiceGiraffe (Cascaded Captions) | 36.7 | Overall accuracy (%) over the 1,500 four-option VoiceGiraffe | 9.1 |
| VoiceGiraffe (Cascaded Captions) - Multi-Hop Reasoning | 30.6 | Accuracy (%) on the 500 multi-hop questions (causal alignmen | 9.1 |
| VoiceGiraffe (Cascaded Captions) - Single-Hop Perception | 39.7 | Accuracy (%) on the 1,000 single-hop questions (temporal loc | 9.1 |
Interactive version: theaggregate.ai/model?slug=moss-audio-8b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.