Qwen model lab
Apache-licensed open weightsQwen
Alibaba Cloud’s Qwen team builds multilingual, multimodal and agent-focused models across a wide range of deployable sizes.
Latest tracked open release
Qwen3.6 open family
The Qwen3.6 open family targets agentic coding with a dense 27B model and an efficient 35B-A3B mixture-of-experts model.
- Released
- 22 April 2026
- Licence
- Apache 2.0
- Dense model
- Qwen3.6-27B
- MoE model
- 35B total, 3B active
- Context
- Up to 262K tokens
Why it matters
Where Qwen stands out
- Strong agentic coding
- Efficient dense and MoE options
- Broad local deployment support
Go to the source