Overview
LLM for Mac is a reference for one question: which local model fits which Mac. PC and GPU builds, including VRAM tiers, stay on the sister site llmhardwareguide.com.
Who publishes this
The publication name is LLM for Mac. Pages do not use a personal byline. The repository is AXIA-Enterprises/llmformac. Issues go to that repository.
What a page will and will not claim
A number appears on a page only when a primary source states it. Acceptable sources are Apple spec pages, model cards, and vendor docs such as Ollama, LM Studio, MLX, and llama.cpp. The access date is written next to the citation.
If the source does not state a figure, the page says unknown. That includes speed. This site does not publish tokens per second unless a primary source does, and then only as a dated quotation. The llama.cpp quantize README includes speed columns that do not name Apple silicon. Those speed columns are not repeated here. File size and bits per weight from that README are quoted, with the URL and the date 2026-10-12.
Memory fit is arithmetic from methodology. The formula is shown, and each assumption is labeled as an assumption.
License
Page text is published under CC BY 4.0. Attribute “LLM for Mac (llmformac.com)“.