Qwen3-235B-A22B

Qwen3-235B-A22B, released by Alibaba's Qwen team on April 28, 2025, is an open-weight Mixture-of-Experts model supporting both fast standard generation and slower extended thinking — offering developers the choice between quick responses and deeper reasoning in a single model. Its open-weight release under a permissive license made large-scale MoE architecture with hybrid thinking accessible to a wide range of researchers and organizations who needed an alternative to closed API models for reasoning-intensive tasks.