Downloads of Python Tools That Pull Text Out of PDFs Grew 64.8x Between 2022 and 2026, and AI-Focused PDF Readers Nearly Doubled Their Share

Summary

Picture a home builder who posts a price sheet and a warranty guide as PDFs. For years those files were mostly read by people who clicked a link. Now they are also read by AI pipelines that pull the text out, split it into chunks and feed it to language models that answer customer questions.

That extraction step is done by Python libraries. Some are long-standing PDF tools; others, released since 2023, are built to turn PDFs into Markdown or structured text for AI. Their downloads on the Python Package Index show how fast PDF reading by machines is growing.

In the Q3 2026 sample days, ten PDF extraction packages were downloaded 293,770,398 times, 64.8x the same days of 2022. PyPI overall grew 9.5x. The AI-focused packages grew fastest of all and nearly doubled their share of the group in two years.

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.

Classic PDF packages: pypdf, pypdf2, pdfminer-six, pdfplumber, pymupdf, camelot-py. AI-focused PDF packages, built to convert documents for language models: pymupdf4llm, docling, unstructured, marker-pdf. Classic libraries are also used for non-AI work such as invoices and reports, and some AI-focused packages install classic ones as dependencies.

PDF extraction downloads grew 65x in four years

Quarterly totals for the 21 sampled days of each quarter.

Growth sped up in 2025 and 2026. Total PDF extraction downloads in Q3 2026 were 5.9x those of Q3 2025.

AI-focused PDF readers nearly doubled their share

Download totals by package, Q3 2026 sample days.

The largest AI-focused package in Q3 2026 was pymupdf4llm with 15,242,612 downloads, followed by docling with 8,117,518. Classic libraries still carry most of the volume, partly because AI-focused tools often install them underneath.

Automated pipelines account for a growing share

CI and installer signals for all ten packages combined.

The AI-focused group is more automated than the classic group: 35.5% of its Q3 2026 downloads were CI-flagged against 29.5% for classic libraries, and 84.9% used uv.

What this means

Any PDF you publish can be pulled apart and quoted by an AI tool, often without the context of the page that linked to it. Old price lists, outdated brochures and unlabeled scans can end up shaping how an assistant describes your business. Remove or update stale PDFs, give each one a clear title and first page, and make sure important details also appear as normal page text. A PDF review belongs in any AI search optimization services package, because old files keep getting read long after you forget them.

If you rely on PDFs for menus, price sheets or guides, it is worth checking what AI assistants say about them. Our free AI visibility audit covers this, and PDF cleanup is one of the easier wins in LLM SEO.

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.

Classic PDF libraries are used for many non-AI tasks, so not every download relates to AI.

AI-focused packages often install classic PDF libraries as dependencies, so one install can count in both groups.

Frequently asked questions

Related studies

• 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 AIMCP Downloads Now Exceed 10 Agent Frameworks Combined
• 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 AIThree AI Providers Hold 89.9% of Model SDK Downloads