M4 Max 64GB: every decision model, every small LLM, real work, done.
TECHNICAL DIFFICULTY: 1 OF 4 ยท PLUG IN AND GO
How hard is this to assemble? one box, one cable: it ships whole, plug it in
The build that answers 'is local even worth trying yet?' with yes. 64GB runs the 27B class comfortably and the sub-10B class at hundreds of tokens per second. For the decision-model wave (Jev-style routing, Laya, form fillers) this is more machine than you need, which is exactly the point. It will not run the 200B+ MoE class at anything but tight quantization, and that is fine: those models belong one tier up.
WHO THIS BUILD IS FOR
You want local AI as a weekend project, not a hobby that eats your garage. Small voice assistants, email triage with a decision model, document search, a private chat: all inside the house.
POWER, NOISE, AND THE ROOM IT LIVES IN
About 140W under load, quieter than the laptop it replaces, one power cable like a lamp. No UPS, no thermal planning, no case fans to pick.
NEXT: THE $5k TIER
This is the tier where local AI becomes boring, which is the highest compliment. The box is silent, the answers are fast, and the models are big enough that you stop asking what a model can do and start asking what you want done. Your writing assistant, your code reviewer, your document librarian, your form-filler: all local, all day, on hardware that uses less power than the monitor next to it.
See the $5k tier →