Latest GPT-4o "analyzes" for a while then says, "done!" without producing output

With the latest version GPT-4o has become almost useless as it constantly tries to run code on its own (somehow) and fails to actually write any code in the UI. I keep reminding it to actually write something, and telling it not to even attempt to execute code as it doesn’t have the data, conda env, or anything else it needs. It’s completely absurd. Am I the only one experiencing this? I’m a long way into a project/chat and it’s become borderline unusable. To be honest, it was doing this before, on occasion, but it’s become way more frequent.

Incredibly aggravating.