What's up with the Plugins model today?

Anybody else getting bizarre, fictional, completely inaccurate behavior from the plugins model today? Check this out:


In every instance, Web Requests returned accurate, current results, and no matter how specific I was in my instructions, GPT returned either semi-accurate results (demonstrating a clear ability to understand the content being returned), or entirely fictional results, and could not seem to follow the instruction.

Any ideas?