AI 101

Apple Silicon (Mac)

Also known as: M1 · M2 · M3 · M4 · Apple GPU

Local

Apple Silicon (M1/M2/M3/M4) Macs are a popular local-LLM platform thanks to fast unified memory and strong Metal support.

  • Unified memory shared by CPU + GPU.
  • 16–64+ GB configurations let bigger models fit.
  • Excellent price/performance for on-device inference.

The "big memory" route: a 32–64 GB Mac can run 13B–70B models that no consumer Windows GPU touches — at the cost of lower raw speed.