Anyway; I y33ted ChatGPT into the sun because it also doesn't work if you don't have a mobile phone number to give it "for authentication".

Installed Llama.cpp and grabbed one of the smaller Gemma-4 models, so I have something that runs locally on an open source LLM that isn't terrible. It's not cutting edge either, but it's enough to be helpful.

And because it runs on my laptop; I know where the electricity comes from to power it.