Guides
Every guide
Start with the memory guide, try the calculator, then choose a setup. Technical claims are sourced; benchmark performance has not been independently tested.
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?
Model weights, attention cache, GB versus GiB, and a configurable memory budget — with explicit assumptions and a worked example.
Setup guideOllama on Mac in 2026: setup, and why it now runs on MLX
Understand the documented MLX backend, confirm model support, and check the installed runtime before running a model.
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?
Start with your existing Mac, estimate the complete memory allocation, and verify current configuration prices before considering an upgrade.