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,...
| Field | Per token (USD) | Per million |
|---|---|---|
| prompt | 0.00000014 | $0.140 |
| completion | 0.0000008 | $0.800 |
| input_cache_read | 0.00000005 | $0.050 |