How parsing works, and its limits
Updated 2026-11-17
Two paths
- Source path: when a paper has LaTeX source on arXiv, Clipaper parses the source, which gives the most accurate formulas, numbering and cross-references.
- PDF path: other papers are read page by page from the PDF and split into paragraphs, equations, figures, tables and references.
- If the source can’t be parsed, Clipaper switches to the PDF path at no extra cost. The paper’s parsing details show which path was used.
Page and paper limits
- Free plan: 3 papers a month, up to 40 pages each.
- Members: credits per page, up to 80 pages per paper, up to 3 papers at the same time.
- Pages are counted from the whole PDF, including the cover, references and appendices.
How long it takes
Most papers are done within a minute or two; long papers take longer. You can read the original PDF in the meantime, and we let you know when it’s ready.
What can’t be parsed
- Password- or DRM-protected PDFs and damaged files can’t be parsed; you’ll see this when importing.
- Tables or formulas that can’t be recognized reliably are shown as a screenshot of the original and marked “Not recognized”, rather than as wrong content.
- If parsing fails, the free parse or credits are refunded automatically.
Reporting problems and re-parsing
If a card is split wrongly or a formula or table is off, right-click the card and choose “Report a parsing problem”, and re-parse if you like. Each paper has 2 free re-parses, and your notes and layout move to the new result.
Didn’t find an answer? Choose “Help & feedback” in the avatar menu in the app, or email support@clipaper.app.