Convert PDF to Markdown
A PDF’s text, readable by Git, wikis and AI agents.
Free, no account, no sign-up. Choose your PDF, download its Markdown. A PDF does not hold a document’s structure: Paginel reconstructs it, approximately. Proofread the result. The file is not kept.
Processed in memory, then discarded. 20 conversions every 10 minutes, 100 a day.
Convert a PDF to Markdown
- Choose the file A PDF of up to 10 MB; or drop it on the area.
- Choose where images go Embedded in the
.mdin base64 (a single file), or stored in animages/folder, all in a.ziparchive. - Download and proofread The Markdown bears your file’s name. Nothing is kept on the server. Read it through: headings, lists and tables are the first things to check.
PDF to Markdown: what is carried over
- The text, with bold, italic and web and email links; headings (
#,##…) and bulleted or numbered lists, nested. - For a tagged PDF (Paginel’s, and Word’s with its default settings): headings, lists, tables, links and images, from the file’s structure; tables are written as GitHub tables (GFM).
- JPEG and pixel (8-bit) images, in base64, or numbered in the order of the text in the
images/folder. - At the top of the file: the language, the margins, the header and the footer when the PDF repeats some.
What the conversion does not do (yet)
- A PDF is not a document: it places letters and images on pages, with no paragraphs, headings, lists or tables. Paginel reconstructs them: the result is approximate and must be proofread.
- A PDF without tags: paragraphs, headings (from size, then bold) and lists (bullets, “1.”) are guessed, and tables are not recognised: their rows become text.
- Several columns: the text of a page is read left to right across its whole width, so columns get mixed up line by line.
- A scanned PDF (page images): no optical character recognition; the pages are carried over as images, with no text.
- A password-protected PDF is refused. Text that is not horizontal on the page is not read; equations and floating text boxes are plain placed text.
- The colour, font and size of a passage are written as HTML
<span>when they differ from the body text: depending on the PDF, the Markdown may be full of them. A PDF has no styles: whatever is neither a heading nor a list is directly formatted text. - Vector drawings (charts) are lost; JPEG 2000, CCITT and JBIG2 images are not carried over.
The Paginel editor’s engine
Markdown is Paginel’s pivot format: the language of its Markdown view and of its AI agents. The PDF is read by a reader written for Paginel, which produces a document of the same model as the editor’s; the Markdown is written by the code that writes your .md files in the editor. It is the newest of our conversions: it keeps improving, and the limits above are today’s.
Going further
Write and edit together in Paginel: open your .docx and .md files in the browser, everyone writes in their own section, and the document exports to Word, Markdown or PDF. Free during the beta.
Try the demo Create an account Create my account My documents
Frequently asked questions
Is it really free?
Yes, with no account, no sign-up and no ads. To keep the service available to everyone, one IP address may run 20 conversions every 10 minutes and 100 a day, two at a time.
Are my files kept?
No. The file is processed in memory for the duration of the conversion, then discarded: it is never stored nor added to any documents, even when you are signed in. Only your IP address is kept in memory, for one day at most, to count conversions. Nothing is passed to a third party. See the privacy policy.
Are headings and lists kept?
In part, and it is a reconstruction: a PDF holds no headings or lists, only text placed on pages. Paginel finds the structure from the file’s tags when there are some (Paginel’s PDFs, and Word’s with its default settings), otherwise it guesses headings (size, then bold) and lists (bullets, “1.”). Proofread the result.
Images in base64 or in a folder?
In base64, you get a single .md file, easy to send but heavier. With the box ticked, a .zip archive: the Markdown and an images/ folder, linked by relative paths, handy for Git, a wiki or a static site.
My PDF is a scan: will the text be recognised?
No. Paginel does no optical character recognition (OCR): the pages of a scanned PDF are only images, so they are carried over as images, with no text to edit. A PDF with no text and no image to carry over is refused. For a scan, first run an OCR tool that adds the text to the PDF.
My PDF is password-protected: can I convert it?
No: an encrypted or password-protected PDF is refused. Remove the protection first (with the software that created the file, or a PDF reader), then try again.
Why did my tables turn into text?
A PDF does not know what a table is: it places text and rules on the page. When the PDF is tagged (Paginel’s, and Word’s with its default settings), its tables are carried over as tables, to be proofread: some cells may be misplaced. Without tags, Paginel recognises no table at all: their rows become lines of text, and the table has to be rebuilt.
Why are there <code><span></code> tags in my Markdown?
Markdown has no colour, font or size: when a passage has one that differs from the body text, it is written as HTML. A PDF often has some, and the resulting Markdown may be full of them; delete them if you do not need them.
How faithful is the conversion?
Only one figure is measured: on six real Word documents run through Paginel’s PDF, then read back as Markdown (an internal test, run locally), 98 to 99% of the words were found again, with the same numbers of list items and tables, give or take one. That figure says nothing about the layout, and PDFs from other tools (LaTeX, LibreOffice, browsers, Pages, scanners) have not been measured: always proofread the result.
What file size is accepted?
Up to 10 MB. Each conversion runs on its own, with bounded time and memory: a PDF that is too heavy or too complex is refused with a message that says so.
The other free tools
- Word to PDF A .docx to PDF, with its headings, tables, images and footnotes.
- Markdown to PDF A .md file, or a .zip with its images, to a laid-out PDF.
- Word to Markdown A .docx to clean Markdown, images in base64 or in a folder.
- PDF to Word A PDF to an editable .docx: an approximate reconstruction, to be proofread.
- All tools