Qwen 3: Alibaba ships an open-weight family from 0.6B to 235B with native thinking
In one sentence Alibaba ships Qwen 3: 8 models from 0.6B to 235B params (2 MoE + 6 dense), all with switchable thinking mode. Apache 2.0 license. Repositions Qwen as the best open weight.
Alibaba (via the Qwen team) ships the third generation of its open-weight model family. It's a huge drop: 8 different models, from the smallest (600M params, runs on a phone) to the largest (235B total, 22B active, frontier-level).
All have a new trait: "thinking mode" can be toggled on and off. You can ask the model to answer fast (no thinking) or to reason at length (thinking on). It's the first open weight with this native flexibility.
Apache 2.0 license — free for commercial use too. 119 languages supported, including Italian. Almost immediately it becomes a base for independent fine-tunes and edge devices (Snapdragon X Elite, Apple Silicon).
Companies
Alibaba
Tools
Qwen3-235B-A22B, Qwen3-30B-A3B, Qwen3-32B, Qwen3-14B, Qwen3-8B, Qwen3-4B, Qwen3-1.7B, Qwen3-0.6B
Tags
Sources