Apple

Mac Mini M4 Pro with 48GB unified memory — a compact local inference powerhouse. Run Llama 3.1 70B Q4 locally.

MacBook Pro M4 Max with 128GB unified memory — run 70B+ models at full precision on a laptop.