AI TCH PH Phil McNeely How I Got a Mac Mini 2018 to Run GPU Inference Over Thunderbolt 3 I run local LLM inference as a hobby that’s gotten slightly out of hand. I have a couple of inference machines already (a full write-up of…