Qwen 3 4B (Non-reasoning): benchmark results
Provider: Alibaba. Released 2025-04-29. Access: Open.
Aggregate Rating 1435 ± 10, rank #1329 of 2201 rated models, from 379 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SRD Reference - MATH-500 (non-thinking) | 84.2 | pass@1 (%) | 100 |
| EuroEval Ukrainian Summarization - LR SUM UK | 31.56 | chrF3++ (%) | 97.2 |
| EuroEval Albanian Summarization - LR SUM SQ | 34.22 | chrF3++ (%) | 95.4 |
| EuroEval Catalan Summarization - Dacsa CA | 38.84 | chrF3++ (%) | 92.9 |
| BAGEL - Xeno-canto Harmonic Structure | 40.3 | Accuracy (%, times 100) on the 427 Xeno-canto questions on h | 91.7 |
| EuroEval Serbian Summarization - LR SUM SR | 30.35 | chrF3++ (%) | 91.2 |
| EuroEval Icelandic Summarization - RRN | 37.83 | chrF3++ (%) | 88.3 |
| EuroEval Ukrainian NLU - Cross Domain UK Reviews | 61.43 | Matthews correlation coefficient (%) | 88 |
| EuroEval Bosnian Summarization - LR SUM BS | 30.91 | chrF3++ (%) | 87.8 |
| EuroEval Italian Summarization - Ilpost SUM | 38.3 | chrF3++ (%) | 87 |
| EuroEval English Summarization - CNN Dailymail | 43.2 | chrF3++ (%) | 86.7 |
| EuroEval Portuguese Summarization - Publico | 33.3 | chrF3++ (%) | 86.6 |
Interactive version: theaggregate.ai/model?slug=qwen-3-4b-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-12.