Qwen: Qwen3 235B A22B Instruct 2507

Qwen3-235B-A22B-Instruct-2507 is a multilingual, instruction-tuned mixture-of-experts language model based on the Qwen3-235B architecture, with 22B active parameters per forward pass. It is optimized for general-purpose text generation, including instruction following,...

• Context window: 262,144 tokens
• Input: text
• Output: text
• Pricing: $0.09/M input tokens, $0.1/M output tokens

View on OpenRouter. Model data sourced from OpenRouter.