O que esta página responde
The Qwen3.5 122B-A10B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. In terms of
- Alibaba Cloud · Qwen · qwen/qwen3.5-122b-a10b
- text+image+video->text · rota de modelo da China
- 262.144 context · US$ 0,26 input