Atlas

Models

← All models

Qwen3 0.6B Base

Alibaba · 2025-04-29 · 596M parameters

Qwen3-0.6B-Base is the pre-trained (not instruction-tuned) 0.6B dense checkpoint of Alibaba's Qwen3 family, released April 29, 2025 under Apache 2.0 alongside the post-trained Qwen3-0.6B. Qwen describes a three-stage pre-training on about 36 trillion tokens in 119 languages, ending with long-context training at 32,768 tokens. It has 596,049,920 distinct parameters, counting the tied input and output embedding matrix once.

Benchmark scores

BenchmarkScore
BBH (BIG-Bench Hard)41.5
Capability89.6
GSM8K59.6
MATH32.4
MGSM31.0
MMLU52.8
MMLU Pro24.7
MMLU-Redux51.3
SuperGPQA15.0
Loading Atlas data…