Find Citations
Paste your research text to discover relevant papers from 3 million+ arXiv articles
0/4,000
Ctrl + Enter to search
How it works
Stop hunting Google Scholar for hours.
Underleaf turns a paragraph from your draft into a ranked list of relevant papers in seconds. No boolean queries, no keyword gymnastics.
01
Paste your text
Drop in a paragraph, abstract, or topic description from your draft. At least a sentence or two so the model has signal to match against.
02
Semantic vector search
Your text is embedded and matched against 3M+ arXiv papers by meaning, not keywords. Surfaces relevant work even when the wording differs.
03
Cite in one click
Read the abstract, open the paper on arXiv, and copy a ready-to-paste BibTeX entry directly into your .bib file.
Why semantic search
The right paper, even when you don't know its terminology.
Beyond keywords
Traditional search misses papers that describe the same idea in different language. Embeddings find them.
3M+ arXiv corpus
Covers cs, math, stat, physics, and more. Refreshed regularly to include recent preprints.
Built for LaTeX
Every result ships with a clean BibTeX entry. Drop it straight into your .bib file and \cite it from your manuscript.
The full toolkit
Log in to Underleaf for the full research toolkit.
Free searches here are capped at 3. Sign in to keep searching, plus convert PDFs to LaTeX, images and handwriting to LaTeX, tables and equations to LaTeX, and write with an AI-native LaTeX editor that knows your project.
No credit card required.
Frequently asked
About the citation finder.
How does the AI citation finder work?
Your text is converted into a vector that captures its meaning, then matched against an index of 3 million+ arXiv papers. Results are ranked by semantic similarity, so they match meaning rather than keyword overlap. You find relevant work even when the wording is completely different from your draft.
Is it really free?
Yes. You get 3 free searches on this page with no signup. For additional use plus the rest of the Underleaf research toolkit (PDF-to-LaTeX, image-to-LaTeX, AI LaTeX editor, and more), create a free account in the app.
What papers does it search?
The arXiv corpus, with over 3 million preprints across computer science, mathematics, statistics, physics, quantitative biology, and economics. Updated regularly to include recent submissions.
What kind of text should I paste?
A paragraph from your draft, an abstract, or a topic description works best. The model performs better with more context, so a sentence or two is the minimum. A full paragraph is ideal.
Does it return BibTeX?
Every result includes a one-click BibTeX entry that you can paste straight into your .bib file. The entry uses the standard archivePrefix=arXiv format with title, authors, year, and the eprint identifier.
How is this different from Google Scholar?
Google Scholar searches by keyword and citation graph. Semantic search matches by meaning, so it finds relevant papers even when they describe the same idea in completely different language. That helps when you don't yet know the exact terminology used in the literature.
Will it find recent papers?
Yes. The arXiv index is refreshed regularly so new preprints are included. For papers outside arXiv (journal-only publications, books, theses), pair this with a tool like Google Scholar or your library's database.