Qwen 3 235B A22B (Non-reasoning): benchmark results
Provider: Alibaba. Released 2025-04-29. Access: Open.
Aggregate Rating 1547 ± 15, rank #890 of 2168 rated models, from 144 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| QuarkMedBench (1,000-Word Limit) - Highlight | 45.11 | Highlight score rate (%): share of the Highlight rubric poin | 100 |
| pfgen-bench - Completion Mode - Fluency | 0.9 | Fluency Score | 99.7 |
| pfgen-bench - Completion Mode - Helpfulness | 0.59 | Helpfulness Score | 99.7 |
| pfgen-bench - Completion Mode - Score | 0.81 | pfgen Score (mean of three) | 99.7 |
| pfgen-bench - Completion Mode - Truthfulness | 0.95 | Truthfulness Score | 99.7 |
| QuarkMedBench (1,000-Word Limit) - Important | 69.62 | Important score rate (%): share of the Important rubric poin | 90.9 |
| BFI-Adapt | 28.7 | BFI-Adapt composite (%; mean over the 27 event-trait pairs w | 84.6 |
| PodBench - Instruction Following | 93.64 | Instruction Following (0-100; judge-derived checklist, items | 83.3 |
| VinCLAT (Catalan) | 9.1 | Average Accuracy (%) | 82.4 |
| QuarkMedBench (1,000-Word Limit) | 67.71 | Base score (%): the rubric score rate, weighting Essential, | 81.8 |
| PodBench | 76.02 | PodBench Score (0-100; mean of instruction following and scr | 77.8 |
| PodBench - Script Quality | 58.39 | Podcast Script Quality (0-100 rubric: depth 45, structure 30 | 77.8 |
Interactive version: theaggregate.ai/model?slug=qwen-3-235b-a22b-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-20.