PDF to Markdown Converter

1. Drop a PDF here, or tap Choose a PDF to browse for one.
2. If some pages are scans, pick the OCR language that matches the document.
3. Wait for the text to appear. Pages with a text layer are read directly; scanned pages go through OCR, which takes longer.
4. Copy the Markdown, download it as a .md file, or save a ZIP that also holds the page images and the pictures found in the PDF.

No account, no sign-up and no daily limit: convert as many PDFs as you need, one after another.

A lot of people convert a PDF to Markdown before handing it to an AI assistant. Markdown is plain text with light marks for headings, lists and emphasis, so you can see exactly what the model will read, cut the parts it doesn't need, and paste the rest into ChatGPT, Claude or another LLM tool.

Working from text also makes it easier to quote the source correctly. When the model answers, you can search the Markdown for the passage it used and check it against the PDF.

A .md file is just as handy outside AI tools. It drops into a notes app such as Obsidian, sits in a Git repository next to your code, and stays searchable years later.

Some PDFs are only pictures of pages, with no text inside. For those pages the converter runs OCR (optical character recognition) with tesseract.js, right in your browser. The language data is served from this site, not from a third party.

OCR covers 20 languages. In Latin or Cyrillic script: English, Spanish, French, German, Italian, Portuguese, Polish, Turkish, Indonesian, Vietnamese and Russian. In Chinese, Japanese, Korean, Indic or Thai script: Chinese (Simplified and Traditional), Japanese, Korean, Hindi, Bengali and Thai. Written right to left: Arabic, Persian and Urdu.

OCR reads what it sees, and it can misread. Check names, numbers and tables against the original before you rely on them.

• Copy: puts the Markdown on your clipboard, ready to paste into a chat or a document.
• Download .md: saves a single Markdown file you can open in any text editor.
• ZIP: the Markdown plus an image of every page and the pictures found inside the PDF.

The ZIP matters when a chart or a diagram carries part of the meaning. Text alone loses it, but with the images saved next to the Markdown you can attach them too.

• Use the original digital PDF when you have it. Pages with a text layer are read directly, with no OCR involved.
• For scans, pick the document's language first. OCR is more accurate when it knows which language to expect.
• Tidy the text before an AI sees it. Delete repeated headers, footers and page numbers if they appear, so the model doesn't treat them as content.
• Split very long documents. A PDF can have up to 200 pages here, but a chat model may not accept that much text in one message.

A PDF can be up to 128 MB and 200 pages long. Everything is processed on your device: reading the text layer, the OCR and writing the Markdown all happen in your browser. Nothing is posted to a server, since the site has no upload endpoint.

That makes it reasonable to use on reports, contracts or research notes you would not paste into an unknown website. The only place the text goes is wherever you paste it next.

Need pictures of the pages rather than text? Save them with the PDF to PNG converter or the PDF to JPG converter. To mark up or fill in the PDF first, use the online PDF editor.