ChatPDF vs Paperlayer

ChatPDF lets users summarize documents and ask questions about their contents using conversational AI. Answers include citations that point to references in a PDF, and users can organize files in folders or discuss multiple PDFs in one conversation. It accepts PDFs in any language and supports conversations in any language; listed file types also include Word, PowerPoint, Markdown and text. Sharing links can be created, with access permissions managed or revoked. ChatPDF says documents are encrypted during transfer and storage and can be deleted by the user. Queries are routed dynamically between GPT-4o and GPT-4o-mini. The service works in a browser on any device and is listed for Android, iOS, Linux, macOS and Windows. Its backend API accepts uploaded PDFs or public URLs and lets users query them to build custom PDF chatbots. The API documentation limits each PDF to 2,000 pages or 32 MB and describes the chat API as stateless. The Free plan costs nothing and allows two documents per day.

Paperlayer is a web-based AI reading layer for research papers, pairing summaries and contextual margin notes with evidence in the original PDF. You can find public arXiv papers by title or author, enter an arXiv ID or paste an arXiv link. Summaries cover the paper’s problem, method, results and caveats, while highlights mark claims, methods, results, problems and caveats with adjustable density. Margin notes explain terms, equations and figures beside the text. Summaries and notes support Korean, Japanese, Chinese, Spanish, French and German, and J and K keys navigate through a paper. Paper chat is marked “Soon.” The public demo covers public arXiv papers only and accepts no private or unpublished uploads. Paperlayer says live processing sends paper and request content to OpenAI over encryption without intentionally including account, cookie or anonymous identifiers.