I did not find how to enable reasoning with the proxy. The model I use supports it but it does not seem to return any reasoning content. Did I miss something ? This is the curl command: curl -N http://localhost:16666/v1/chat/completions \ -H "Content-Type: application/json" \ -d '{ "model": "gemma4-31b", "messages": [ {"role": "user", "content": "Write a haiku about technology"} ], "stream": true, "chat_template_kwargs": {"enable_thinking": true} }'