Turn a PDF into clean Markdown
Convert a PDF into clean Markdown with its headings and paragraphs preserved. Useful for pasting a document into a notes app, committing it to a repository, or handing it to a language model that reads structure far better than it reads a wall of text.
PDF to Markdown
font-size heading detection
Loading tool…
How to convert PDF to Markdown
- Add your PDF. Drag the file in. Nothing is uploaded — the document is read in this tab.
- Choose a page range (optional). Convert the whole document or narrow it to the section you need.
- Convert. Text is grouped into blocks, and blocks set noticeably larger than the body become headings.
- Copy or download. Copy the Markdown straight to your clipboard, or save it as a .md file.
How headings are found
A PDF has no headings. It has glyphs at sizes, and a heading is simply text that happens to be bigger. So the conversion works out the document's own body size first — the size the bulk of its text is set in — and treats blocks well above that as headings, one level for each step up.
Judging size relative to the document rather than against fixed point values matters: a paper set in 9pt and a report set in 13pt disagree completely about what counts as large. A length test runs alongside it, because a whole paragraph set slightly big is emphasis rather than a section title, and promoting it would wreck the outline the markup is meant to produce.
Why Markdown rather than plain text
Extract Text gives you every word as one flat stream, which is the right answer when the words are all you want. Markdown keeps the shape of the document: sections stay distinguishable, lists stay lists, and the page boundaries are marked.
That structure matters most when something else has to read the result. A model summarising a forty-page report does noticeably better when it can see where the sections begin, and a notes app can fold the document by heading rather than showing an undifferentiated block.
Scans have nothing to convert
If the PDF came from a scanner, its pages are images and contain no text at all. The tool says so rather than returning an empty file. Run OCR PDF over it first to add a real text layer, then convert.
Frequently asked questions
Is the file uploaded to convert it?
No. Reading the PDF and generating the Markdown both happen inside your browser. The document never leaves your device.
Are tables converted to Markdown tables?
No. Table content comes through as text in reading order. Rebuilding rows and columns from glyph coordinates is guesswork that produces confidently wrong tables often enough not to be worth it.
Are images included?
No — the output is text only. To get the pictures out of a PDF, use Extract Images.
Why are some headings the wrong level?
Levels come from relative font size, so a document that uses size inconsistently will produce an imperfect hierarchy. It is usually a small edit to fix, and it is far quicker than marking up the whole document by hand.
What are the horizontal rules in the output?
Page boundaries from the original PDF. They keep the original pagination visible without inventing headings that were never in the document.
Can I convert just part of a long report?
Yes. Set a page range before converting and only those pages are read.
Is this good for feeding a PDF to an LLM?
That is one of the main reasons it exists. Markdown keeps the section structure that helps a model navigate a long document, and it costs fewer tokens than the equivalent HTML.