76.1% of Vector Database Client Downloads Came From Automated CI Pipelines in 2026, Up From 15.1% in 2024

Summary

Picture a law firm's AI intake assistant that answers questions about practice areas by searching a store of text chunks taken from the firm's own website. That store is a vector database, and it is where web content goes after it is scraped and converted for AI.

Python apps connect to vector databases through client libraries published on the Python Package Index. Their download counts give a rough view of how many projects are building retrieval systems, the setup that lets an AI answer from specific documents or web pages.

In the Q3 2026 sample days, eight client packages were downloaded 130,914,428 times. Most of those downloads came from automated test pipelines: 76.1% were CI-flagged, up from 15.1% two years earlier. Downloads outside CI grew too, but at about the pace of Python overall.

What we measured

We queried the public PyPI download log in Google BigQuery (bigquery-public-data.pypi.file_downloads), which records one row for every file downloaded from the Python Package Index. To keep the query affordable we sampled the first seven days of every month from January 2022 through September 2026, 399 days in total, and summed downloads by month, package, country, CI flag, installer and weekday versus weekend into one working table. A quarter in this study therefore means 21 sampled days (days 1 to 7 of each of its three months), not a full quarter. We removed downloads made by mirroring and proxy tools (bandersnatch, Nexus, Artifactory, devpi), which copy packages in bulk rather than install them. To control for the growth of PyPI itself we also pulled total daily download counts for the whole index from the table's partition metadata for the same sampled days.

We tracked eight client packages for seven vector stores: weaviate-client, qdrant-client, faiss-cpu, chromadb, lancedb, pinecone and pinecone-client (both Pinecone), and pymilvus. Shares are download shares only, not market share or a measure of quality. Because CI downloads are concentrated in some packages, we show shares both for all downloads and for non-CI downloads.

Three in four vector database downloads run in CI

CI-flagged and non-CI downloads, Q3 sample days.

Most of the two-year growth came from CI. CI-flagged downloads rose 54.8x, while non-CI downloads rose 3.1x.

Download share by vector store

Download counts and shares for Q3 2026 sample days. The non-CI column removes downloads made inside test pipelines.

The order changes once CI downloads are set aside. Among non-CI downloads, Qdrant (24.6%), FAISS (21.4%) and Chroma (15.6%) led. These are adoption signals only.

Total downloads grew sharply from a small base

Total downloads per quarter (21 sampled days).

The series jumped in Q2 2025 and again in 2026, both times driven mainly by CI downloads. US IP addresses accounted for 90.2% of Q3 2026 downloads, which mostly reflects where cloud servers and CI runners sit.

What this means

When an AI system stores your web pages in a vector database, it saves them as chunks of text and pulls back the chunks that best match a question. Pages built from short, self-contained sections, each with a clear heading and a direct answer, tend to produce cleaner chunks than long pages where key facts are buried mid-paragraph. Dates, prices and service areas written in plain text survive the process better than those in images.

If you want AI systems to retrieve the right facts about your business, structure key pages for it. A free AI visibility audit shows which pages to restructure first, so your LLM SEO effort goes where retrieval systems actually look.

Limits

A download is not a user. One developer, one Docker build or one CI job can download the same package many times, and a single busy pipeline can move a package's numbers sharply from month to month.

Figures cover the first seven days of each month only (21 days per quarter), so they are sample totals, not full-month or full-quarter totals. Growth rates compare like-for-like sample periods.

PyPI as a whole grew from 12.1 billion downloads in the Q3 2022 sample days to 115.0 billion in the Q3 2026 sample days (9.5x), so part of every package's growth is the index growing. We report share of all PyPI downloads where it matters.

PyPI-wide downloads in the September 2026 sample week were 38.1% lower than in the August 2026 sample week (28.7 billion vs 46.4 billion), so September figures for individual packages should be read with that drop in mind.

The CI flag is only populated in this dataset from March 2024, and it only catches environments that announce themselves as CI. Automated downloads that do not set a CI variable are counted as non-CI.

Country codes come from the downloading IP address. Cloud servers sit mostly in the United States, so country splits describe where machines run, not where developers live.

Vector database clients are used for many retrieval tasks beyond web content, including internal documents and product data.

A few packages account for most CI downloads, so totals and shares can swing with individual projects' test setups.

faiss-cpu is a library rather than a hosted database client, but it serves the same retrieval role.

Frequently asked questions

Related studies

• Downloads of the tools that read websites for AIMCP Downloads Now Exceed 10 Agent Frameworks Combined
• Downloads of the tools that read websites for AIAI Agent Framework Downloads Topped 108M in One Week
• Downloads of the tools that read websites for AIWeb-to-Text Tool Downloads Grew 71x Since 2022
• Downloads of the tools that read websites for AIPDF-to-Text Downloads for AI Rose 65x in Four Years