It even runs large language models

Anything can run an LLM, it's just a matter of how many parameters and how slow it is. People have put them on retro computers.

And people wonder why I didn't pay out my ass to upgrade my vidya card. All my monitors are 1080p. What a waste that would have been.