PDF to Markdown
Convert PDF to Markdown in your browser. Headings, lists, tables, links, code and two-column pages come out as GitHub-flavoured Markdown, with a preview, a copy button and a .md download.
Runs in your browser — nothing is uploadedFree trial: 1 run of PDF to Markdown every day. After that, 2 credits per run.
Turn a PDF into clean Markdown you can edit, compare or give to an AI model
Open a PDF and get GitHub-flavoured Markdown with its headings, paragraphs, bulleted and numbered lists, tables, links, bold and italic text and code blocks. Tagged PDFs are read from their structure tags, the same way a screen reader reads them. Untagged files are rebuilt from the layout, so two-column pages, repeated headers and footers, page numbers and words broken across lines are dealt with too.
Everything happens in your browser. The Markdown appears next to a rendered preview, so you can check it before you copy it or download a .md file.
What PDF to Markdown gives you
Keeps the outline
Headings come from the tags, the bookmarks or the type sizes, so the Markdown keeps the shape of the document instead of turning it into one long block of text.
Tables come out as tables
Text that lines up in columns over several rows becomes a Markdown pipe table, and a cell that wraps onto a second line is joined back together.
Two columns, read in order
Pages set in two columns are read down the left column and then the right one, instead of straight across both.
Page clutter removed
Headers, footers and page numbers that repeat from page to page are left out, and a word split with a hyphen at the end of a line is put back together.
Links and code survive
Web links stay links, and text set in a monospaced font becomes inline code or a fenced code block that keeps its indentation.
How to use PDF to Markdown
Open the PDF
Choose a file. If it is password protected, type the password to open it; it is not stored.
Pick pages and options
Convert every page or a range such as 1-5, 8, and choose whether to remove headers, rebuild tables, mark page breaks or add front matter.
Check, then copy or download
Compare the Markdown with the preview, then copy it or save it as a .md file.
Other names for this tool
People also look for this as PDF to MD, convert PDF to Markdown, PDF to Markdown converter, PDF to text with formatting and PDF to Markdown for LLMs. It is the same tool on this page.
Looking for the opposite job? Markdown to PDF.
Limitations
- No text recognition: scanned pages without a text layer produce no text.
- Up to 500 pages and 100 MB per file. Forms, magazine layouts and text inside images may need tidying by hand.
Questions about PDF to Markdown
Is my PDF uploaded?
No. The PDF is opened by pdf.js in your browser and the Markdown is written there as well; nothing from the document is sent to a server.
Does it work on scanned PDFs?
Only pages with a text layer can be converted. Pages that are pictures of paper come out empty, and the tool lists them so you can run text recognition on them first.
How are tables found?
Tagged PDFs keep the tables they tag. In other files, text that lines up in the same columns over at least three rows is rebuilt as a Markdown table.
Why is a page empty in the Markdown?
That page has no text layer. It is a scanned image or a photo of paper, so there are no characters to read, and the tool lists those pages after converting. Run them through text recognition (OCR) first, then convert the result.
How are headings found in a PDF without tags?
From three clues, in order: the bookmarks, when the file has them; text set clearly larger than the body text; and short bold lines on their own. The largest size becomes a level 1 heading, the next a level 2, and so on.
Can I use the Markdown with ChatGPT or another AI model?
Yes. Markdown keeps headings, lists and tables as plain text that language models handle well, and the conversion leaves out the repeated headers and broken lines that clutter text copied straight from a PDF. Turn on Mark page breaks if you want answers to cite page numbers.
Why did a table come out as plain lines?
Tables are rebuilt only when at least three rows line up in the same columns, so short or irregular tables stay as lines of text, and a table that is a picture has no text at all. In a tagged PDF, tagged tables are kept as tables. Table rebuilding can also be switched off.
Does it work with password-protected PDFs?
Yes, if you know the password. The tool asks for it, opens the file with it in your browser and forgets it when you leave the page. A PDF that only restricts copying opens without a password.