npm and PyPI Package Stats

Download counts and release facts for npm and PyPI packages, in one dataset. One row per package.

Useful for a briefing or a dashboard. Download counts include automated installs, so treat a spike as a question, not a verdict.

Who it is for

Developer relations teams comparing adoption, open source maintainers watching their own packages, investors reading a library before a conversation, and people building tools that need the numbers in a dataset.

What you get

• One row per package, with the ecosystem and the name you asked for.
• Download counts for the latest published day, week and month. Those windows can lag by a day or two, and they are not the same as the date range below.
• A daily series for the dates you set, and the total of that series.
• The latest version and the day it was released.
• Release cadence: how many releases fell in the last year, the median days between them, the gap since the previous release, and how many fell inside your dates.
• The licence, when the package publishes a short identifier.
• The number of dependencies declared on the latest version. Python extras are not counted.
• The repository link, when it points at GitHub, GitLab, Bitbucket, Codeberg or SourceHut. A personal site is left empty.
• A trend against the previous period, when you turn that comparison on. The trend is a direction (up, down or flat) and a percent change. It is left empty when either period was not published in full.
• A note on the row when a figure was missing. The other figures on that row are still saved.

Tutorial

1. Set Packages. Each item needs a name and an ecosystem, npm or pypi. A scoped npm name is written @scope/name.
2. Leave From and To empty for the 30 days ending yesterday, or set both. The range can be at most 180 days.
3. Leave Comparison on None, or choose Previous period to compare that range with the same number of days just before it.
4. Leave Max packages at 25 unless you want a shorter run. The ceiling is 40.
5. Run the Actor. Open the Packages view. The OUTPUT record says how many rows were saved and names any package that was not found.
6. Schedule it if you want a standing feed. Packages that were not saved are not charged.

Example input

{
  "packages": [
    { "name": "express", "ecosystem": "npm" },
    { "name": "requests", "ecosystem": "pypi" }
  ],
  "dateFrom": "2026-10-01",
  "dateTo": "2026-10-02",
  "comparisonMode": "previous-period",
  "maxPackages": 25
}

Example output

{
  "ecosystem": "npm",
  "name": "express",
  "latestVersion": "4.21.2",
  "latestReleaseDate": "2025-12-01",
  "downloadsDaily": 1800000,
  "downloadsWeekly": 12000000,
  "downloadsMonthly": 50000000,
  "seriesStart": "2026-10-01",
  "seriesEnd": "2026-10-02",
  "seriesDownloads": 3600000,
  "series": [
    { "date": "2026-10-01", "downloads": 1800000 },
    { "date": "2026-10-02", "downloads": 1800000 }
  ],
  "releasesLastYear": 6,
  "medianDaysBetweenReleases": 48,
  "daysSincePreviousRelease": 30,
  "releasesInRange": 0,
  "licence": "MIT",
  "dependencyCount": 31,
  "repositoryUrl": "https://github.com/expressjs/express",
  "trendMode": "previous-period",
  "trendDirection": "up",
  "trendChangePercent": 12.5,
  "trendCurrentDownloads": 3600000,
  "trendPreviousDownloads": 3200000,
  "partial": false,
  "note": ""
}

The numbers above show the shape of a row. A real run fills them from the sources below.

Pricing

You pay a small fee when the run starts, then a fee for each package row that is saved. The daily series is part of that row. The dataset itself is not charged again.

What	Price
Run start	$0.00005
Package stats row	$0.001

Packages past Max packages, and names that are not published, are not charged. A row that is saved with a missing figure is charged, and the note says what was missing.

Examples

• Ten package rows: 10 × $0.001 = $0.010, plus the $0.00005 start fee.
• A run that saves nothing: $0.00005.

Use Max packages and Apify's maximum cost per run to cap a single run. The pricing tab on the store is the price you are charged if it ever differs.

Scheduling and integrations

Save a task and schedule it when you want a weekly check. The dataset can go to Google Sheets, Slack, email, Make or Zapier, or you can read it from the Apify API. The rows are small enough to pass to an AI agent over MCP.

Sources and licences

npm figures come from the public npm registry and the public npm download counts. The npm Open Source Terms allow commercial use, and they allow reading the public registry through its public interface. They also forbid an unreasonable volume of requests. This Actor caps each run and waits when a source asks it to slow down. This Actor is not affiliated with npm.

The licence on a row is the identifier the package publishes. It is not a licence for the download counts. Download counts describe installs, including continuous integration, not a count of people.

PyPI project facts come from the public PyPI interface, under the PyPI Terms of Service. Those terms forbid using the interface to sell personal information. Python download totals and the daily series come from PyPI Stats, a public project of the Python Software Foundation. The project asks callers not to pull every package, because the figures update once a day and the service is small. The daily series covers about half a year, so a Python row can be shorter than the dates you asked for. The project code is under the Apache License 2.0. This Actor is not affiliated with the Python Software Foundation.

Known mirror copies are left out of the Python daily series.

No personal data

This Actor does not collect personal data.

• It asks for package names, a date range and a comparison mode. It does not ask for account names or contact details.
• Author, maintainer and contributor fields are dropped, including names and emails.
• Raw responses are not stored.
• Logs and error messages do not include personal data or the addresses that were called.

FAQ

Why do the daily count and the series differ? Daily, weekly and monthly are the latest windows the source has published. The series is the date range you set. A package can be busy last month and quiet in the week you asked about.

What if a name is wrong? That package is skipped. The run summary names it. Other packages are still saved. If nothing could be saved, the run stops with that message and no package rows are charged.

Why is a Python series missing days? PyPI Stats keeps about half a year of daily points, and a day that was not published is left out. The note on the row says when the series is incomplete. The trend is empty unless both periods are complete.

Where do the figures come from? See Sources and licences above. For a binding claim about a project, read the project itself.

Next additions

• More package ecosystems.
• More signals about each project.

Feedback

Missing a field? Open an issue on the Actor's Issues tab.