Qwen 3 0.6B: 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.
Aggregate Rating 1309 ± 7, rank #1459 of 1656 rated models, from 861 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Supra-microBench - Coherence and Common Sense | 50.1 | Chance-normalized score (0–100), v0.2.0 | 100 |
| Supra-microBench - Context Use and Reading | 64.03 | Chance-normalized score (0–100), v0.2.0 | 100 |
| Supra-microBench - Elementary Reasoning | 77.1 | Chance-normalized score (0–100), v0.2.0 | 100 |
| Supra-microBench - Index | 69.82 | Chance-normalized score (0–100), v0.2.0 | 100 |
| Supra-microBench - World Knowledge | 87.94 | Chance-normalized score (0–100), v0.2.0 | 100 |
| TurkBench - Toxicity | 100 | LLM-as-a-judge score (0-100) | 100 |
| Polar (US) - Economic Axis | 77.74 | Economic-axis ICAT (0-100), the StereoSet idealized context | 97.3 |
| Supra-microBench - Lexicon and Semantics | 79.42 | Chance-normalized score (0–100), v0.2.0 | 94.1 |
| Supra-microBench - Linguistic Competence | 59.82 | Chance-normalized score (0–100), v0.2.0 | 94.1 |
| CSSBench - Privacy Leakage | 6.67 | Attack success rate (%; share of obfuscated malicious privac | 88.9 |
| PreAct-Bench - Daily Life | 66.7 | Prefix Foresight F1 with the full prefix (PFF(100), binary F | 87.5 |
| Syslog Severity Classification - RAG | 88.12 | Accuracy on 9,355 held-out journalctl logs, exact syslog sev | 87.5 |
Interactive version: theaggregate.ai/model?slug=qwen-3-0-6b · How It Works · Data refreshed daily, snapshot 2026-10-20.