picomatch
Blazing fast and accurate glob matcher written in JavaScript, with no dependencies and full support for standard and ex…
npm says
697,111,787
per week
Every tarball fetch: mirrors, scanners, CI and people.
real installs
high confidence≈ 697M
per week
Mirror spikes and crawlers removed. CI and dependency installs still count, so this is not a head count.
noise under 1% of npm's count, so nearly all of it counts. noise real installs
Why this estimate
With JavaScript on, this lists what the filter found for this package.
Most used major version: 4, the latest one, with 60% of last week's downloads above the crawler floor. Most used single version: 4.0.7.
Snapshot from the daily build of Nov 9, 2026, last 7 days of npm data. With JavaScript on, this page recomputes it live from the npm API.
How many people use picomatch?
npm counted 697,111,787 downloads of picomatch in the last 7 days. Nobody can count people exactly, npm included: it does not record who downloaded what. After removing mirror spikes and crawlers, about 697M look like real installs. At this scale most of them come from CI, builds and other packages that depend on picomatch, so the number of people behind them is far smaller.
Is picomatch still maintained?
Most likely yes, going by its releases. The latest version, 4.0.7, came out on Aug 24, 2026. The last 12 months saw 6 stable releases, typically 77 days apart. It was first published on Nov 5, 2018. npm lists 4 maintainers.
Is picomatch still used?
Downloads with publish spikes removed are up 47% over 3 months and up 278% over 11 months. For scale, the median of the 384 packages we track, which lean towards today's most downloaded, is up 164% over 11 months, so this one is ahead of that. Last week, 60% of downloads above the crawler floor were on major version 4, the latest one; the most used single version was 4.0.7, with 30%. Weekdays get 1.5× the downloads of weekends, which fits people at work, though CI follows the same rhythm. Crawlers account for about 1,687 downloads a week, roughly 48 per version.
Monthly downloads of picomatch
The busiest month was Oct 2026, with about 89,589,987 downloads a day once publish spikes are removed; the quietest was Dec 2025 (the holiday season), with about 22,682,490.
| month | npm downloads | publish spikes removed | real installs | vs month before |
|---|---|---|---|---|
| Oct 2026 | 2,239,749,680 | 0 | 2.24B | up 14% |
| Sep 2026 | 2,127,922,442 | 0 | 2.13B | up 20% |
| Aug 2026 | 2,030,301,337 | 0 | 2.03B | about flat |
| Jul 2026 | 1,988,167,478 | 0 | 1.99B | up 11% |
| Jun 2026 | 1,730,389,372 | 0 | 1.73B | up 13% |
| May 2026 | 1,588,855,753 | 0 | 1.59B | about flat |
| Apr 2026 | 1,482,338,250 | 0 | 1.48B | up 19% |
| Mar 2026 | 1,283,129,794 | 0 | 1.28B | up 29% |
| Feb 2026 | 897,203,477 | 0 | 897M | up 21% |
| Jan 2026 | 822,878,513 | 0 | 823M | up 17% |
| Dec 2025 | 703,157,196 | 0 | 703M | — |
Whole months through Oct 31, 2026. Publish spikes are found day by day and summed per month. Real installs apply this week's crawler share to each month's downloads (the upper end also keeps the removed spikes), so they follow npm's curve rather than measure it. The change compares spike-free downloads per day.
How the estimate works
- Cut the spike every new version gets from mirrors in its first days.
- Subtract the flat floor robots download from every version each week.
- Show what is left as a range, with a confidence level. It still includes CI and dependency installs, so it is not a count of people.