pdfGPT is one of the original open-source “chat with your PDF” tools: upload a PDF and ask questions about its contents in natural language, answered by a GPT model grounded in the document. It has around 7,200 stars on GitHub and is still maintained in 2026.
Long before chat-with-documents became a standard product category, pdfGPT demonstrated the pattern that hundreds of later tools would copy: extract the text of a PDF, index it, and let a large language model answer questions using the document as context. The result is a simple, effective workflow for researchers, students and professionals who need to interrogate contracts, papers, manuals and reports without reading every page. The project is deliberately lightweight and hackable, written in Python under the MIT license, and a public demo space shows the experience in the browser. Its enduring popularity — thousands of stars and a long tail of forks — reflects how well it captures the essential chat-with-PDF experience without unnecessary complexity. It is completely free to self-host with your own API key.
pdfGPT is completely free and open source under the MIT license. You run it yourself and only pay for any LLM API usage involved in answering questions.
pdfGPT is for students, researchers and professionals who want to query long documents conversationally, and for developers looking for a minimal, proven starting point for building document-chat applications.