Skip to content

GLM 5.3 Flash reasons before answering again

GLM 5.3 Flash sometimes answered without showing its reasoning: at High, the level it used by default, the model barely reasoned. Max is now the recommended level, so it thinks before answering every request. It takes a few seconds longer and uses a little more of your usage per answer.

If you picked another level in the Effort row, it stays. Admins can cap it in Organization → Models with Maximum effort. More about effort in Writing a request and about your organization’s models in Which models your organization uses.