Last updated: October 2, 2026
pdfmux is an open-source Python library that runs entirely on your machine. It processes PDF files locally — no data is sent to any server unless you explicitly configure an optional cloud extractor.
None. The pdfmux library does not collect, transmit, or store any personal data, usage data, or telemetry. Your PDFs stay on your machine.
This website is a static site hosted on GitHub Pages. We do not use cookies, analytics, or tracking scripts. GitHub may collect standard server logs (IP address, browser type) as part of hosting — see GitHub's privacy statement for details.
If you install pdfmux[llm] and configure a Gemini API key, pdfmux will send page images to Google's Gemini API for extraction. This is opt-in and requires your explicit API key configuration. The data handling for that request is governed by Google's API terms.
When you install pdfmux via pip install pdfmux, your request goes through PyPI. We do not have access to any data from pip installs.
pdfmux offers optional integrations with LangChain and LlamaIndex. These integrations run locally and do not transmit data on their own. Any data transmission depends on how you configure those frameworks in your application.
The pdfmux tool inside ChatGPT is different from the library: it runs on our server, so your PDF does leave your machine. It is operated by Drumworks Ventures FZ LLC (United Arab Emirates).
What we receive. The PDF you upload in ChatGPT, or the PDF at a link you give, plus an anonymous user identifier and conversation identifier created by OpenAI. We do not receive your name, email address or the rest of your conversation.
What we do with it. The PDF is read in an isolated process with no network access, its tables are returned to ChatGPT, and the PDF is deleted as soon as the call finishes. The Excel file we create is available through a private link for 15 minutes and then deleted.
What we keep. A usage record for each call: the anonymous identifiers in hashed form, the number of pages, tables and rows, whether rows reconciled, and timing. It never contains the contents of your document. These records are used for fair-use limits and to understand usage, and are not shared or sold.
OpenAI handles your ChatGPT conversation under its own privacy policy. To ask about or delete records linked to an identifier you can supply, contact us as below.
Questions about this policy? Open an issue on GitHub.