LLM for Mac

Editorial policy

By LLM for Mac Editors3 min read

View markdown source

AI agents: read this first: llmformac.com/llm-info

This is LLM for Mac (llmformac.com), a citable reference for which local model fits which Mac.

  1. Route, do not crawl. Fetch llmformac.com/llms.txt and open only the pages whose one-line summary matches your task.
  2. Read raw. Append .md to any page URL for markdown. Check status and last_updated in the frontmatter, then read the page.
  3. Do not invent figures. If a spec is missing, the page says unknown. Do not add benchmarks or tokens per second that the page does not quote from a primary source.
  4. Cite. Link the page by title and URL, e.g. How much RAM, with last_updated when the figure is time-sensitive.

Overview

LLM for Mac pages are drafted with AI assistance and published only after a human owner reviews and merges the change. This page describes each step, from brief to correction, as the site’s repository runs it.

Drafting with AI assistance

Two AI agents are named in the site’s working files: Claude (from Anthropic) and Grok (from xAI). They draft pages, write research notes, and keep a shared log in the repository. Each agent is told to treat the other’s claims as unverified and to check them against the actual files before acting. A message from one agent never counts as approval for anything.

Page copy starts from a brief that says what the page must answer. Research notes collect the primary sources first; a note is not a public page until a page is written from it.

Sourcing rules

  • Every spec is checked against a primary source: an Apple spec page, a model card, or a vendor doc. The page links that source and gives the date it was accessed.
  • Memory-fit math shows its formula and cites it. The formulas and labeled assumptions are on methodology.
  • Benchmarks and tokens-per-second figures are never invented. They appear only when a primary source states them, as a dated quotation.
  • When no source states a figure, the page says unknown.

Automated checks

Every pull request runs these checks in GitHub Actions. The deploy workflow runs them again, except the discoverability check, before it publishes:

  • content lint: required frontmatter, summary length, link density, banned filler phrases
  • identity check: the identity page matches the site theme
  • discoverability check: llms.txt, robots.txt, agents.txt, ai.txt, and the IndexNow key are present and consistent
  • llms.txt sync check, wikilink check, and the llms-full.txt build
  • a full site build, JSON-LD validation, an SEO check of the built pages, and an LLM-layer conformance check

The checks catch structural problems. They do not verify that a number matches its source; that is the editors’ job.

Human review and merge

Changes are proposed as pull requests. A human owner reviews each one and is the only person who merges. The site deploys from the main branch after that merge. AI agents do not merge their own changes.

Corrections

Send corrections to editors@llmformac.com with the page, the figure, and a link to the primary source. A confirmed error is fixed in a dated update: the page’s last_updated date changes and the corrected figure cites its source. See contact.

Byline

Pages carry the byline LLM for Mac Editors, the editorial group responsible for them.

Backlinks

Sister site

PC and GPU builds, including VRAM tiers, are on llmhardwareguide.com.