Qwen 3.5 0.8B: benchmark results
Alibaba's Qwen3.5 0.8B, a small-scale post-trained checkpoint in the new Qwen3.5 family that unifies vision-language understanding, intended mainly for prototyping and task-specific fine-tuning given its size. Provider: Alibaba. Released 2026-03-02. Access: Open.
Aggregate Rating 1332 ± 8, rank #1424 of 1656 rated models, from 646 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MERA Code - CodeLinterEval | 100 | pass@1 (%) | 100 |
| PRISM-VLM - Sycophancy Resistance | 81 | Per-axis mean score (%; share of turn-0-correct answers held | 97.6 |
| EuroEval Bosnian Summarization - LR SUM BS | 30.23 | chrF3++ (%) | 79.4 |
| MERA Industrial - ruTXTAgroBench - Crop cultivation technologies | 46.1 | Score (%) | 78.6 |
| EuroEval Polish Summarization - PSC | 23.59 | chrF3++ (%) | 77.6 |
| EuroEval Spanish Summarization - Mlsum ES | 27.29 | chrF3++ (%) | 72.8 |
| MERA Industrial - ruTXTAgroBench - Crop production | 34.2 | Score (%) | 70.9 |
| AA-Omniscience Hallucination Rate | 61.35 | Hallucination Rate (%) | 70.5 |
| EuroEval Ukrainian Summarization - LR SUM UK | 28.94 | chrF3++ (%) | 69.7 |
| EuroEval Bulgarian NLU - Cinexio | 41.68 | Matthews correlation coefficient (%) | 67.8 |
| EuroEval Portuguese Summarization - Publico | 32.42 | chrF3++ (%) | 66.1 |
| EuroEval Portuguese NLU - ScaLA PT | 18.35 | Matthews correlation coefficient (%) | 64.7 |
Interactive version: theaggregate.ai/model?slug=qwen-3-5-0-8b · How It Works · Data refreshed daily, snapshot 2026-10-20.