Academic PDFs to clean Markdown.
Equations, tables & citations intact.
Drag-drop any paper and download a .md file with equations as LaTeX, tables as real Markdown tables, and references as numbered footnotes — ready for ChatGPT, Claude, Obsidian, or your RAG pipeline.
No API key · No account required · Result in ~30–90 seconds
[garbled table █████ ███]
References: [1, 2, 3, 4]
— equations broken, table mangled
| Model | BLEU |
|-------|------|
See[^1], [^2], [^3].
Three steps, under a minute
Pay & drop your PDF
Checkout in seconds, then drag your paper onto the upload page. No account, no setup.
We convert it
A layout-aware engine extracts equations as LaTeX, rebuilds tables, and maps citations to footnotes.
Download the .md
Get a clean Markdown file, ready to paste into your notes, an LLM, or a RAG pipeline.
What you actually get
Valid LaTeX equations
Inline and display math come out as real LaTeX — not garbled Unicode or dropped symbols.
Tables that survive
Multi-column academic tables become clean Markdown pipe tables, not merged or broken rows.
Citations as footnotes
In-text references and the bibliography are converted to numbered [^n] Markdown footnotes.
Zero setup
No API key, no SDK, no credit math. Drag, drop, download — straight from the browser.
LLM-ready output
Works out of the box with ChatGPT, Claude, Obsidian, and retrieval pipelines.
Private by default
Your file is processed, served back, then deleted. Never used for training.
Simple pricing
Pay per paper, or go unlimited if you convert a lot.
Per paper
$9
One paper, one payment.
- ✓ Pay only for what you convert
- ✓ No account required
- ✓ Instant downloadable .md file
Unlimited monthly
$19/mo
Cancel anytime.
- ✓ Unlimited conversions
- ✓ Best for active research
- ✓ Pays for itself after 3 papers
Secure checkout via Stripe · Cancel subscriptions anytime
Questions
What types of equations does it handle?
A layout-aware parser targets valid LaTeX for inline and display math, with an OCR fallback for equation-heavy or image-based PDFs. arXiv-style STEM papers with a text layer convert most reliably; scanned papers with image-rendered equations may have lower fidelity.
How are citations converted?
In-text citations and bibliography entries in numbered (STEM/arXiv) style are detected and mapped to numbered Markdown [^n] footnotes. APA/MLA/Chicago support is on the roadmap.
How is this different from LlamaParse or Mathpix?
LlamaParse and Mathpix are API/SDK products with credit-based or setup-fee billing aimed at developers. PaperToMD is a drag-and-drop tool that handles equations, tables, and citations together — no code, no credit math, flat pricing.
What happens to my PDF?
Your file is held only long enough to process and serve your download, then deleted automatically. It is never used for training or shared with third parties. See our Privacy Policy.
How fast is it, and is there a free tier?
Most papers convert in about 30–90 seconds depending on length and complexity, and download the moment they're ready. Pricing is $9 per paper or $19/month unlimited — there's no free tier.