PyPI package downloads API for weekly volume

Track exact PyPI project names with weekly download volumes and period growth through the python source on Trends MCP.

Live data as of 2026-08-11

The PyPI package downloads API on Trends MCP is the python source on get_growth and get_time_series. Pass an exact PyPI project name and receive weekly download volume plus a 0-100 index. From the August 11 2026 pull, pandas showed recent_volume 200,739,859 on 2026-08-03 with +19.81% volume growth over 30 days, while requests led the sample at 437,267,339 weekly downloads. There is no PyPI top-packages board; every call needs a project name. Free tier: 100 requests per month.

What PyPI package download data shows right now

From the Trends MCP python growth pulls of August 11, 2026 (series recent_date 2026-08-03), ten widely used projects returned complete volume rows:

ProjectWeekly downloads (2026-08-03)30D volume growth12M volume growth
requests437,267,339+13.17%+138.12%
numpy300,393,248+19.09%+154.14%
pydantic288,882,699+21.14%+205.25%
httpx213,969,742+23.56%+274.60%
pandas200,739,859+19.81%+118.24%
fastapi150,473,727+29.74%+439.84%
openai114,718,603+34.69%+402.17%
langchain62,414,541-16.98%+230.98%
anthropic50,378,710+32.54%+757.61%
torch22,637,359-0.37%+59.42%

Normalized index snapshots on the same recent date: pandas, requests, numpy, openai, pydantic, httpx, and fastapi each sat at 100.0; langchain was 79.4; torch was 92.9; anthropic was 100.0. Over 6 months, openai volume grew +203.17% (37,839,388 to 114,718,603) while anthropic grew +315.49% (12,125,020 to 50,378,710). langchain was the clear near-term cooler: -16.98% over 30 days after still being up +230.98% over 12 months.

The exact API calls for PyPI volumes

get_growth is the fast path when the question is period change. get_time_series is the path when a chart or custom math needs every weekly point.

{
  "mode": "get_growth",
  "source": "python",
  "keyword": "pandas",
  "percent_growth": ["30D", "3M", "6M", "12M"]
}
{
  "mode": "get_time_series",
  "source": "python",
  "keyword": "fastapi"
}

Keyword rules are strict. pandas works; a mistaken display label does not. Scoped npm-style names are irrelevant here. One source plus one keyword counts as one request, and every period inside a single percent_growth array stays inside that request. Auth and client setup are covered on the docs page.

Reading volume growth without fooling yourself

Absolute weekly downloads matter more than the percent when packages sit at different scales. requests at 437.3M weekly downloads moving +13.17% over 30 days is a larger unit shift than anthropic at 50.4M moving +32.54% over the same window, even though the percent is smaller.

Short windows can disagree with long ones. langchain fell 16.98% over 30 days (75,180,838 to 62,414,541) while still posting +230.98% over 12 months. torch was nearly flat over 30 days at -0.37% volume growth (22,721,750 to 22,637,359) with a calmer +59.42% over 12 months. Prefer 30D and 3M for operational monitors; keep 12M for structural adoption stories.

Index values are relative inside each series. A 100.0 on pandas and a 100.0 on fastapi do not mean equal downloads. Always print recent_volume beside the index when the decision depends on size.

Where PyPI data fits beside npm and GitHub

Python package adoption answers a different question from GitHub trending ranks. GitHub shows which repositories are hot today; PyPI shows which libraries are actually installing this week. Teams that already watch JavaScript packages on the npm trends API can run the same growth windows on python for the Python half of a stack. Open-source scouts who start from the GitHub Trending Repos API can verify whether a newly charting Python project has matching PyPI volume after the star spike.

A useful triangulation for AI tooling in this pull: PyPI openai at 114,718,603 weekly downloads (+34.69% over 30 days) and PyPI anthropic at 50,378,710 (+32.54% over 30 days) both expanded while langchain cooled over the same month. That split is invisible on a GitHub board alone.

Limits and honest caveats

The python source has no live top board, so cold-start discovery needs an external name list. Keywords must match PyPI project names exactly; import paths such as cv2 will not resolve to opencv-python. History length varies by package: anthropic returned 183 series points in metadata while pandas returned 261. The newest week in this pull was 2026-08-03 even though the API call ran on 2026-08-11, so freshness labels should use the series date.

Growth percentages on smaller projects swing harder than on requests or numpy. Pair every percent with recent_volume and baseline_volume. Trends MCP does not invent reasons for a download spike; if the cause is a CI mirror, a dependency pin, or a release train, that explanation has to come from outside this API.

Pricing for package monitors

Free covers 100 requests per month. Starter is $19 per month for 1,000 requests, Pro is $49 for 5,000, and Business is $199 for 25,000. Watching 20 packages once per day is about 600 requests per month on growth-only jobs, which fits Starter. Add time-series pulls only for packages that need charts. Confirm the live numbers on pricing before automating.

Common questions

Use the exact PyPI project name, the same string that appears on pypi.org. Examples that returned data on August 11 2026: pandas, requests, numpy, openai, pydantic, httpx, fastapi, anthropic, torch, and langchain. Display names and import paths that differ from the project name will miss.
No. get_top_trends does not expose a PyPI chart. Discovery starts from a known project name, then get_growth or get_time_series on source python returns weekly volume and normalized index values.
On the August 11 2026 pull, successful python growth rows used recent_date 2026-08-03. That is the latest complete week in the series, not the calendar day of the API call. Quote both dates when publishing numbers.
Both return weekly download volumes for exact package names. npm uses JavaScript package names; python uses PyPI project names. On 2026-08-03, requests had recent_volume 437,267,339 while pandas had 200,739,859. Compare normalized indexes when the absolute scales differ.