PDF to Markdown
Turn a PDF into clean Markdown with its headings and lists intact — then see roughly how many tokens you just saved. Handing an AI model a PDF costs several times more than the same content as text, because each page gets processed as an image too.
Drop your file here
or click to browse
Turn this off if your document uses large text for emphasis rather than for headings.
- ProcessingOn your device
- Uploaded to a serverNothing
- InternetNot needed
Don't take our word for it. Turn off your Wi‑Fi, then use this tool. It will work exactly the same, which is only possible because your file never goes anywhere.
Worth knowing
Headings are worked out from font size, which is how a PDF actually stores them — there's no explicit heading marker to read. It's right on most documents and occasionally over-eager on ones that use large text for emphasis. Skim the output before feeding it to anything important.
Token counts are estimates, not exact. They use ~4 characters per token and ~1,600 tokens per page image, which is the right ballpark across current models but will vary by tokeniser.
Why Markdown rather than plain text
Plain text throws away the structure of a document. Markdown keeps it — headings stay headings, lists stay lists, emphasis survives — while remaining readable and costing a fraction of what the original PDF costs in tokens.
That structure is what makes the difference when the destination is an AI model. A heading tells the model where a section begins; a list tells it the items are parallel. Strip that out and the model has to infer it from spacing, which it does imperfectly and at length.
More on this: why converting first costs fewer tokens and gives better answers.
Questions people ask
Why not just upload the PDF to ChatGPT?
You can, and it works — it just costs a lot more. To read a PDF reliably these tools generally process every page as an image as well as pulling the text out, and images are expensive. Converting once and reusing the Markdown avoids paying that on every question.
How much does this actually save?
On a text-heavy document, usually 80-90%. The tool shows you the before and after for your specific file rather than making you take that on faith.
Does it work on scanned documents?
No. A scan has no text in it, only a picture of text, so there's nothing to convert. That needs OCR, which isn't built yet.
Can I use this for Notion or Obsidian?
Yes — that's the same output. Markdown is what both use natively, so you can paste it straight in with headings and lists already formatted.