Qwen 3 Omni 30B A3B: benchmark results
Provider: Alibaba. Released 2025-09-22.
Aggregate Rating 1613 ± 11, rank #901 of 10916 rated models, from 360 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Audio2Tool - Direct Commands | 92.4 | Exact match (%): share of queries whose predicted tool or or | 100 |
| Audio2Tool - Implicit Reasoning | 11.3 | Exact match (%): share of queries whose predicted tool or or | 100 |
| Audio2Tool - Intent Blending | 12.4 | Exact match (%): share of queries whose predicted tool or or | 100 |
| Audio2Tool - Parametric Commands | 15.6 | Exact match (%): share of queries whose predicted tool or or | 100 |
| AudioProcessBench - Cross-Modal Binding Errors | 79.4 | Type-conditioned PRMScore (0-100): detection of steps with c | 100 |
| BASS - Artist Attribution | 63.66 | Exact-match accuracy (%) on the 64 artist attribution questi | 100 |
| BASS - Artist Duration | 9.78 | Exact-match accuracy (%) on the 327 artist duration question | 100 |
| BASS - Artist Localization | 9.38 | Exact-match accuracy (%) on the 247 artist localization ques | 100 |
| CBT-Audio (Audio Only) | 0.79 | Mean absolute error on the 1-5 distress-intensity scale agai | 100 |
| E3mo-Bench (Expressed) - Video-Pair Perception (Yes-or-No) | 66.22 | Accuracy (%; yes-or-no questions comparing the emotion, vale | 100 |
| ESCUCHA | 74.4 | Accuracy (%; all 1,000 questions: 900 multiple-choice items | 100 |
| ESCUCHA - Audio Instruction Following | 88 | Pass rate (%; 100 open-ended items whose instruction (keywor | 100 |
Interactive version: theaggregate.ai/model?slug=qwen-3-omni-30b-a3b · How It Works · Data refreshed daily, snapshot 2026-11-08.