Hey, thanks for the report with the screenshot (TheThe and then nothing). I see what’s happening.
What you’re describing isn’t intended behavior, and it’s not related to your settings. GLM 5.2 (high and max) is currently cutting off the response right after the thinking stage. I reproduced it and sent it to the team. I can’t share an exact timeline yet, but I’ll reply in the thread when there’s an update.
Until it’s fixed, as a workaround I recommend switching to another model, for example Claude Sonnet 5 or GPT-5.6. Those return full responses.
About charges for failed turns, the team handling that is at [email protected]. Email them and they’ll look into your specific billing situation.
Let me know if the issue shows up anywhere else after you switch models.
