Convert PDF to Markdown on your Mac

Drag a PDF into Milldown and get clean Markdown out — the text in reading order, page furniture stripped. Nothing is uploaded: conversion runs entirely on your Mac.

How to convert a PDF to Markdown

1. Open Milldown and drop your PDF anywhere in the window.Or right-click the file in Finder → Services → Convert to Markdown.Or right-click the file in Explorer → Convert to Markdown with Milldown.
2. The conversion runs locally — a 200-page PDF takes under 15 seconds.The token badge shows exactly how much smaller the Markdown is.
3. Copy the result to your clipboard, save it as a .md file, or clean it up first.The built-in Clean Up pass strips repeated page headers, footers, and page numbers.

What the conversion preserves

• Text and reading order — paragraphs come across in the order they are meant to be read, from the PDF's text layer. A PDF stores no heading information, so section titles arrive as ordinary lines.
• Tables, with a caveat — a PDF stores positioned text rather than table structure. Simple tables usually come across as rows you can tidy; dense or multi-column pages can interleave. Worth a look before you rely on one.
• Multi-page documents — handled as one continuous document, with the option to strip per-page boilerplate.

Embedded images aren't carried into the Markdown (it's a text format) — for image-heavy PDFs, keep the original alongside the converted text.

Before and after, on a real document

From arXiv paper 2506.16035 — the public research paper in our benchmark corpus — at a page boundary, where raw PDF extraction is at its worst:

…with specialized ap-
proaches for contextualizing tabular
data in RAG systems [Allu et al.,
2024], though challenges

2

\fremain for tables spanning multiple
pages. Modern document conversion
toolkits like Docling [Li-
vathinos et al., 2025] have provided…

…with specialized approaches for
contextualizing tabular data in RAG
systems [Allu et al., 2024], though
challenges remain for tables spanning
multiple pages. Modern document
conversion toolkits like Docling
[Livathinos et al., 2025] have
provided…

The stranded page number and form feed split that sentence mid-clause — and this junk repeats at every one of the paper's 21 page boundaries. Clean Up strips standalone page numbers and repeated furniture, revertibly. Full arithmetic for this exact file is on the benchmarks page: 14,816 Markdown tokens against the 50,386 Anthropic reports for a native upload.

Scanned PDF? Covered.

If your PDF has no embedded text (a scan or a photographed document), Milldown detects it automatically and runs on-device OCR using Apple's Vision framework. See the dedicated guide: scanned PDF to Markdown.

From the command line

milldown report.pdf -o report.md

Milldown installs an optional CLI (app menu → Install Command-Line Tool) with full option parity — handy for scripts and CI.

Further reading: Measured token savings by file type · Preparing PDFs for ChatGPT · Fixing “context window exceeded”