Could you tell me what exactly this issue is? I’ve encountered it twice already.
I’ve been having this issue today a few times as well. Most of the time GLM 5.2 works good and fast, but on a few occasions its either very slow, reconnecting all the time, or is indeed unable to reach the model provider
constantly all day long it is dropping.
Hey, thanks for the report. This message (“Unable to reach the model provider”) shows up when our backend can’t reach the upstream model provider. It’s usually a temporary network issue or a 5xx error on the provider side after the automatic retries are used up. It’s not an issue with your setup, GLM 5.2 is working as a native model for you.
For your specific request 5a5e7e66-519a-41f9-bf30-e5c4dc763685, it looks like it failed on the provider side before it got to a completion. GLM 5.2 is a popular new model right now, and its endpoint sometimes has error spikes under load. That’s why you’re seeing intermittent “reconnecting” and “unable to reach” like @dirk-mk and @notorioustech mentioned.
What you can do:
- Hit Resume, it will continue the same chat and you won’t lose work.
- If it keeps happening, wait a couple minutes and try again.
- If it’s unstable, switch to a different model for now.
Related discussion of the same issue: Cursor upstream model services experiencing degradation. If the drops keep happening even after switching models, let me know and share a couple fresh Request IDs, and we can dig in.