Ali Tungyi Laboratory open sources Qwen3.6-35B-A3B model.

On April 16, the AliComm Yishi Lab announced the open-source release of Qwen3.6-35B-A3B. It is described as an efficient model that adopts a sparse mixed expert architecture: with a total of 350 billion parameters, only 30 billion parameters are activated during each inference. In terms of intelligent programming, Qwen3.6-35B-A3B significantly surpasses its predecessor, Qwen3.5-35B-A3B, and can compete with larger dense models such as Qwen3.5-27B and Gemma-31B. It is also compatible with mainstream programming assistants such as OpenClaw, Claude Code, and Qwen Code.