For some reasoning models, the Think process is required by default, but sometimes it is not necessary, and a quick response to the question is sufficient;
When making a request, parameters such as “thinking”: { “type”: “enabled” }, need to be added to the ChatComplete API request header to ensure it is turned on or off;
If using the OPENAI Response API, it would be in the form of reasoning={“effort”: “medium”}, etc.;
It is hoped that for large reasoning models like OPENAI, GLM, etc., which optionally support enabling the THINK mode, this parameter can be configured.
Allow users to turn on or off the think/reasoning option or set the level of thinking in the configuration interface for the deployed reasoning service.