Everything you need to know about using PyPI Stats.
Search for any Python package using the search bar. You'll see a detailed dashboard with download trends, version breakdowns, and ecosystem data.
If a package isn't in our database yet, we'll automatically start loading its data from Google BigQuery. This usually takes a few minutes — just refresh the page shortly.
Create an account to save packages to your dashboard, set up alerts, and unlock Pro features like extended history and CSV exports.
Each package dashboard includes:
The health score is a 0–100 rating composed of four equally-weighted components (25 points each):
How stable are daily downloads over 90 days. Low variance scores higher. A package with steady, predictable downloads is a safer dependency.
Total downloads over 90 days. Scored as: 1M+ = 25, 100K+ = 20, 10K+ = 15, 1K+ = 10, otherwise 5.
5 points each for having: a summary, author, license, homepage, and latest version. Well-maintained packages tend to have complete metadata.
Week-over-week download change. 10%+ growth = 25, any growth = 20, minor decline = 15, major decline = 5.
The trending page ranks packages by week-over-week download growth. A package needs at least 1,000 downloads in the past week to qualify — this filters out noise from tiny packages with volatile numbers.
Trending is a great way to discover new libraries gaining traction in the Python ecosystem before they become mainstream.
The compare tool lets you view download trends for multiple packages side-by-side on the same chart.
Free accounts can compare up to 2 packages. Pro accounts can compare up to 10 — useful for evaluating alternatives (e.g., Flask vs FastAPI vs Django) or tracking a suite of related packages.
Alerts notify you via email when something notable happens with a package you're watching. Three alert types are available:
Alerts are checked daily. Create them from your dashboard.
Pro and Enterprise users can enable a weekly digest email that summarizes how their saved packages performed over the past week. It includes download counts and week-over-week percentage changes for each saved package.
Enable it from your dashboard. Digests are sent every Monday at 9:00 AM UTC.
Pro and Enterprise users can toggle “Exclude CI/CD” on package dashboards to filter out downloads from known automated tools and mirrors, including bandersnatch, devpi, Bazel, Nexus, Artifactory, and Homebrew.
Note: most CI/CD pipelines use standard pip or uv to install packages, making them indistinguishable from human installs at the data level. This filter removes what's identifiable, but won't capture all automated traffic.
PyPI Stats provides a public REST API for programmatic access to download data, package metadata, and health scores. See the API documentation for endpoints, rate limits, and authentication details.