Local LLMs thrive on Apple's hardware, and a huge part of it is thanks to MLX.
It's pretty handy for low-power server nodes ...