Verified record · from the answer index

Qwen: Qwen3 235B A22B Thinking 2507

modeltested · static-verifiedscore 20/100signed passport ✓published 457d ago · verified 2026-10-24

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...

Declared tools (2)

text->texttool-calling

model:openrouter:qwen/qwen3-235b-a22b-thinking-2507