Local AI · Apple Silicon
Run serious AI
on your own Mac.
Which Mac, how much memory, which model, which tool. Sourced answers and an honest calculator — written for people who want private, local AI without the guesswork.
Open the memory calculator2026 buying guideStart here
Mac Studio M5 Max vs. M5 Ultra for local AI — and the 512GB question
What Apple's 2026 Mac Studio actually offers for running LLMs locally: memory tiers, bandwidth, prices, and the 512GB configuration listed by Apple.
FundamentalsHow much unified memory do you need for local LLMs?
The honest math behind model size, quantization, KV cache and a configurable memory budget on Apple Silicon — with a worked example.
Setup guideOllama on Mac: setup and MLX support
Install Ollama on an Apple Silicon Mac, run your first model, and check whether it's using the MLX engine introduced as default in v0.40.
Setup guideGetting started with MLX and mlx-lm on Apple Silicon
Apple's open-source MLX framework, step by step: install mlx-lm, run and serve a model, and quantize your own.
ComparisonOllama vs. LM Studio vs. llama.cpp vs. MLX: which should you use?
A plain-English comparison of the four main ways to run local LLMs on a Mac, and who each one is for.
Buying guideThe 2026 buying guide: which Mac for local AI?
How to choose an Apple Silicon Mac for running LLMs locally — memory first, then bandwidth — with published Mac Studio specifications and dated launch prices.