Qwen: Qwen3 235B A22B Thinking 2507

Alibaballm

About

Qwen3-235B-A22B-Thinking-2507 is a high-performance, open-weight Mixture-of-Experts (MoE) language model optimized for complex reasoning tasks. It activates 22B of its 235B parameters per forward pass and natively supports up to 262,144...

Try Qwen: Qwen3 235B A22B Thinking 2507

Test this model in the Sandbase Playground with your own prompts.

Open in Playground

Related Models