Yes, you can disable “thinking” mode in Qwen/Qwen3.6-35B-A3B-FP8 by setting "enable_thinking": false in the chat_template_kwargs field of your API request payload. This works for Qwen3.5/3.6 models in vLLM (source).
Would you like a code example or more detail?
Sources:
Hint: Mention @RunLLM in the post for followups.