State of Mac reports which Mac apps people are getting, from two public sources, and publishes a new issue every week.
Sources
Mac App Store charts
Apple publishes US top charts as RSS/JSON. State of Mac fetches itunes.apple.com/us/rss/topfreemacapps/limit=100/json daily, starting Oct 28, 2026. On Oct 28, 2026 it listed 98 apps. We also fetch toppaidmacapps and topgrossingmacapps daily; on Oct 28, 2026 both returned no entries. We tested more feeds on Oct 28, 2026: Apple's per-category Mac charts (for example genre=12014, Productivity) return no entries, and Apple's newer marketing-tools feeds (rss.marketingtools.apple.com/api/v2/…) cover iPhone and iPad apps, not Mac apps. The category pages under App Store are Apple's overall top free chart filtered by each app's category. App Store ranks are Apple's own and are never combined with the install counts below.
Homebrew install counts
Homebrew is a free tool for installing Mac apps from the command line. Each app it can install is called a cask. Homebrew publishes how many times each cask was installed by people who have not turned its analytics off, at formulae.brew.sh/analytics:
formulae.brew.sh/api/analytics/cask-install/30d.json,90d.jsonand365d.json: install counts for the last 30, 90 and 365 days, ending on the same day.formulae.brew.sh/api/cask.json: each cask's name, description and homepage.
The current edition uses windows ending Oct 28, 2026. The 30-day file lists 13,879 casks and 2,373,781 installs.
Who the install counts describe
Homebrew users lean toward technical users, and only people with analytics left on are counted. The numbers describe that group, not all Mac owners. Apps like Microsoft Office are installed far more often from their own installers or the App Store than through Homebrew. Read install rankings as "among people who install apps with Homebrew". The App Store chart is the better guide to what typical Mac owners download from Apple's store.
Everyday apps and developer tools
Rankings are split in two. Everyday apps (browsers, chat, music and video, office, notes, utilities and so on) lead every page. Developer tools are ranked separately on their own page. A cask counts as a developer tool when it is a command-line tool, driver or plug-in; when it is a pre-release channel such as a beta or nightly build; or when its category is coding agents, editors and IDEs, terminals, containers and virtualization, SDKs and cloud CLIs, or developer tools. A hand-edited list (data/audience.json) moves individual apps either way, for example local AI model runners and remote-desktop servers to developer tools. Desktop apps like ChatGPT and Claude are everyday apps; Codex and Claude Code, which run in a terminal, are developer tools. "Rank" on every page means rank within the app's group. In the current window, everyday apps had 784,423 installs across 3,119 apps and developer tools 1,331,291 across 1,164.
What counts as a Mac app
Rankings include casks from Homebrew's main cask repository, except fonts. Casks from third-party repositories (names with a slash, like owner/tap/name) are excluded because they have no shared metadata, and some of them install Linux software. In the current 30-day window that removes 180,942 third-party installs, 76,367 font installs and 758 installs of casks no longer in Homebrew's index. That leaves 2,115,714 installs across 4,283 casks.
What the numbers mean
- Installs, 30 days: Homebrew's count of install events for the app in the 30-day window. Reinstalls count. Updates of an installed app do not.
- Rank: position by 30-day installs among everyday apps, or among developer tools.
- Share: an app's installs divided by all included installs (everyday apps and developer tools together) in the same period.
- Periods: the three files overlap, so we split them. Days 31–90 = (90-day − 30-day) ÷ 2, per 30 days. Days 91–365 = (365-day − 90-day) ÷ 275 × 30, per 30 days.
- Trend (share change): the percent change between an app's share in days 31–90 and its share in the last 30 days. ▲ 20% means its share is 20% larger. We use share rather than raw installs because the total moved a lot between periods (below).
- Climbing and cooling off: largest share changes among apps with at least 2,000 installs in the last 30 days and 800 per 30 days before (climbing), or at least 2,000 per 30 days before and 1,000 now (cooling off). Apps no longer maintained in Homebrew are left out.
- New: apps with installs in the last 30 days and none in the 335 days before (the 30, 90 and 365-day counts are equal).
The total moved
All included casks averaged 1,864,830 installs per 30 days in days 91–365, 1,492,420 in days 31–90, and 2,115,714 in the last 30 days. We do not know why days 31–90 were lower; it may be usage, Homebrew's data collection, or both. Raw install growth would make almost every app look like it grew, so changes are reported as changes in share.
History
Homebrew publishes rolling totals, not history. We checked the source repository, Homebrew/formulae.brew.sh: analytics are generated when the site is built and have never been committed to git, so past values cannot be rebuilt from it. History therefore comes from:
- Internet Archive captures of the 30-day file. Each capture is a real 30-day window. There are four: windows ending Nov 26 2025, Dec 8 2025, Jun 16 2026 and Jun 27 2026. There are no captures between Dec 8 2025 and Jun 16 2026; charts shade that span.
- Our own snapshots, fetched daily from Oct 28, 2026 onward. Weekly charts get one point per week from here on, and week-over-week comparisons begin with the next issue.
Windows on record:
- Oct 27, 2025 to Nov 26, 2025: 1,482,969 app installs (Internet Archive capture)
- Nov 8, 2025 to Dec 8, 2025: 1,464,118 app installs (Internet Archive capture)
- May 17, 2026 to Jun 16, 2026: 2,067,192 app installs (Internet Archive capture)
- May 28, 2026 to Jun 27, 2026: 2,049,585 app installs (Internet Archive capture)
- Sep 28, 2026 to Oct 28, 2026: 2,115,714 app installs (fetched by State of Mac)
Older captures list fewer casks (8,388 in November 2025 versus 13,879 now), so small apps may be missing from older windows. Category totals over time use today's category for each app.
Categories
Each app is assigned to one of 27 categories. A hand-edited list assigns the most installed apps; the rest are matched by keywords in Homebrew's description. Both files are in the source repository (data/cask-categories.json, data/categories.json) and are editorial. Corrections are welcome.
Guides
Guide pages are built from the same data each week. "Free and open-source apps" are everyday apps whose own Homebrew description contains the word "free" or "open source"; some of them also sell paid plans. "Apps to get on a new MacBook" and "Mac apps for students" list the most installed everyday apps in a fixed set of categories; they do not measure who installed them.
Icons
App icons are the ones each app's homepage declares (its apple-touch-icon, the largest site icon, or /favicon.ico), fetched once and stored. Sometimes that is the publisher's logo rather than the app icon. Apps whose homepage is a code-hosting page, or that declare no icon, show a plain lettered tile. App Store icons are Apple's artwork from the RSS feed.
Updates
A scheduled job fetches the Homebrew files and App Store feeds once a day between 05:17 and 08:17 UTC. A new issue is published weekly. Every fetch identifies itself as stateofmac.com and requests each file once.
Credit and license
Install data is published by the Homebrew project. Homebrew's site code is BSD 2-Clause licensed; the analytics files have no separate license stated, and we republish counts with credit to Homebrew. App Store data is from Apple's public feeds; app names and icons belong to their owners. State of Mac's derived data (rankings, shares, categories) is licensed CC BY 4.0: cite "State of Mac (stateofmac.com), from Homebrew analytics".
Who publishes this
State of Mac is published by Tundra Industries, which also makes Mac apps. Those apps are sold directly and are not on Homebrew, so they never appear in rankings. Pages show them only in boxes labeled "Made by us".