Qwen3-0.6B-Base: benchmark results
Alibaba's Qwen3-0.6B-Base, a dense base model in the Qwen3 series pretrained on 36 trillion tokens across 119 languages through a three-stage pipeline extending context to 32K tokens. Provider: Alibaba. Released 2025-04-29. Access: Open.
Aggregate Rating 1294 ± 34, rank #1464 of 1635 rated models, from 41 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Polar (US) - Economic Axis | 74.91 | Economic-axis ICAT (0-100), the StereoSet idealized context | 91.9 |
| Polar (US) | 78.69 | Total ICAT (0-100), the StereoSet idealized context associat | 81.1 |
| Polar (US) - Sociocultural Axis | 82.47 | Sociocultural-axis ICAT (0-100), the StereoSet idealized con | 64.9 |
| Swallow - Pre-trained English - MATH | 33.4 | Exact Match (%) | 53.1 |
| Swallow - Pre-trained Japanese - MGSM | 30 | Exact Match (%) | 42.9 |
| Swallow - Pre-trained English - HumanEval | 29.5 | Pass@1 (%) | 41.8 |
| Swallow - Pre-trained English - GSM8K | 48.3 | Exact Match (%) | 40.8 |
| Swallow - Pre-trained English - BBH | 40.7 | Exact Match (%) | 34.7 |
| Swallow - Pre-trained Japanese - JCommonsenseQA | 70.5 | Accuracy (%) | 32.7 |
| Swallow - Pre-trained English - Average | 42 | Average Score (%) | 30.6 |
| Swallow - Pre-trained Japanese - JHumanEval | 21.6 | Pass@1 (%) | 30.6 |
| Swallow - Pre-trained English - MMLU | 52.3 | Accuracy (%) | 28.6 |
Interactive version: theaggregate.ai/model?slug=qwen3-0-6b-base · How It Works · Data refreshed daily, snapshot 2026-11-05.