Qwen3 Omni 30B A3B Instruct — benchmark results
Alibaba's open omni-modal 30B A3B MoE: text, image, audio and video in, streaming text and speech out. Provider: Alibaba. Released 2025-09-22. Access: Open.
Unified ELO 1560 ± 29, rank #591 of 1776 rated models, from 46 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Universal Arabic ASR Leaderboard | 30.71 | Average WER (lower is better) | 85.3 |
| Open Universal Arabic ASR Leaderboard - CER | 13.67 | Average CER (lower is better) | 85.3 |
| OCRBench v2 | 61.3 | Average (self-reported) | 69.2 |
| OCRBench-V2 (en) | 61.3 | Score (self-reported) | 68 |
| OCRBench-V2 (zh) | 60 | Score (self-reported) | 64 |
| AA Omniscience - Software Engineering (SWE) - PHP | 28 | Accuracy (%) | 59.5 |
| AA Omniscience - Software Engineering (SWE) - Julia | 16 | Accuracy (%) | 57.8 |
| AudioMC | 24.34 | Score (self-reported) | 55.6 |
| AA LiveCodeBench | 42.22 | Pass@1 (%) | 49.7 |
| AA AIME 2025 | 52.33 | Accuracy (%) | 49.4 |
| SEAL - AudioMultiChallenge | 24.34 | Score | 48 |
| AA Omniscience - Software Engineering (SWE) - Java | 16 | Accuracy (%) | 46.7 |
Interactive version: theaggregate.ai/model?slug=qwen3-omni-30b-a3b-instruct · How the rankings work · Data refreshed daily, snapshot 2026-07-22.