Qwen 3 1.7B: benchmark results
Can switch between direct answers and explicit reasoning within the same conversation, without loading another checkpoint. Training covers multilingual dialogue, code and tool use. Provider: Alibaba. Released 2025-04-29. Access: Open.
Unified ELO 1402 ± 7, rank #1210 of 1639 rated models, from 816 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ActTraitBench | 0.19 | G_KD (Global Knowledge-Decision Gap) (self-reported) | 100 |
| Syslog Severity Classification - Zero-Shot | 33.61 | Accuracy on 9,355 held-out journalctl logs, exact syslog sev | 100 |
| TRACK - Multi-Hop QA (Facts in Context) | 83.6 | Holistic pass (%; multi-hop questions over Wikidata fact cha | 100 |
| EuroEval Ukrainian Summarization - LR SUM UK | 30.85 | Score (%) | 89 |
| EuroEval Catalan Summarization - Dacsa CA | 38.15 | Score (%) | 88.3 |
| EuroEval Italian NLU - ScaLA IT | 42.53 | Linguistic acceptability Score (%) | 88 |
| MERA Code - ruHumanEval | 66.59 | pass@1 (%) | 87.7 |
| Syslog Severity Classification - Few-Shot | 43.3 | Accuracy on 9,355 held-out journalctl logs, exact syslog sev | 87.5 |
| EuroEval Polish Summarization - PSC | 24.52 | Score (%) | 87.1 |
| EuroEval Bosnian Summarization - LR SUM BS | 30.79 | Score (%) | 86.3 |
| EuroEval French NLU - ScaLA FR | 49.2 | Linguistic acceptability Score (%) | 86.2 |
| Kurakurai French LLM - MATH-500 | 65.58 | Accuracy (%) | 85.4 |
Interactive version: theaggregate.ai/model?slug=qwen-3-1-7b · How It Works · Data refreshed daily, snapshot 2026-10-09.