Open Paper

Open Paper is a research workspace for people reading papers and conducting literature reviews. It combines a PDF reader, annotations, and an AI assistant so you can examine answers alongside the papers they draw on.

• Answers with citations: Ask about a paper beside the document, follow citations to the exact supporting passage, or start with an AI-generated brief and suggested questions.
• Highlights and notes: Highlight text, attach comments, and ask the assistant to explain a selected passage, with notes kept in context and available in a Markdown view.
• Research across papers: Group papers into projects and ask questions across your library to get synthesized answers with verifiable citations.
• Structured data extraction: Define the fields you want to compare across a project's papers, collect them in tables with a source for each cell, and export the results as CSV.
• Discovery and library imports: Search open-access literature and add papers to your library, or connect Zotero to import your collection and keep it in sync.

You can also generate audio overviews of papers. The hosted web service requires an account and supports Google or email sign-in.

Open Paper is open source under AGPL-3.0, and you can self-host the full stack. It's built primarily for hosted use, so self-hosting requires more setup. It depends on Postgres, S3-compatible storage, background workers, and LLM API keys; running the application yourself still requires access to an LLM API for AI features.