Alibaba's Qwen Lab Opens Qwen3.6-35B-A3B Model with Sparse MoE Architecture

Gate News message, April 16 — Alibaba's Qwen Lab announced the open-source release of Qwen3.6-35B-A3B, a large language model built on a sparse mixture-of-experts (MoE) architecture. The model features 35 billion total parameters with only 3 billion parameters activated per inference pass.

Qwen3.6-35B-A3B is equipped with agentic programming capabilities and can be integrated into third-party coding assistants, including OpenClaw, Claude Code, and Qwen Code.