How does this compare with a converter that uses an AI model?

A model handles unusual layouts better and can guess at a table's structure. It also means your document is transmitted to whoever runs that model and billed per page. The same file can convert differently on two different days. This one runs locally and returns the same result every time, at the price of doing worse on genuinely messy pages.

Why did my code block come out as a paragraph?

Because the code was not set in a monospace font. Detection keys off the font name reported by the PDF, so code pasted into a document in the same face as the body text is indistinguishable from prose to this tool.

Do I get bold and italic back?

No. Inline emphasis inside a paragraph is not reconstructed. Headings and lists are, along with code blocks and paragraph breaks, and those are the parts carrying the document's structure.

PDFRuns in your browser

PDF to Markdown converter

A converter that reads your document instead of sending it somewhere to be read.

1Documents

Drag PDFs in, paste, or click to choose

A whole folder works too · Ctrl/Cmd+V pastes · on phone this opens your photo library · up to 100 MB per file

Remove repeated headers, footers and page numbers

Mechanism

How this works

Code blocks and lists are detected from the document's own typography. A fragment whose font name identifies a monospace face opens a fenced code block, and the block closes when the font changes back. Bullet characters and numbered prefixes at the start of a line become Markdown list items, with the indentation level taken from the fragment's horizontal position. None of this involves guessing at meaning, which is why it costs nothing to run and returns the same result every time.

Limits

What it cannot do

Code set in a proportional font is invisible to the detector and comes out as an ordinary paragraph.
Inline formatting like bold and italic inside a paragraph is not reconstructed, since it would require reading font weight per fragment and stitching runs back together.
A model-based converter will do better on a messy magazine layout. This one is better on a document that was typeset consistently.

Questions

Questions people actually ask

How does this compare with a converter that uses an AI model?: A model handles unusual layouts better and can guess at a table's structure. It also means your document is transmitted to whoever runs that model and billed per page. The same file can convert differently on two different days. This one runs locally and returns the same result every time, at the price of doing worse on genuinely messy pages.
Why did my code block come out as a paragraph?: Because the code was not set in a monospace font. Detection keys off the font name reported by the PDF, so code pasted into a document in the same face as the body text is indistinguishable from prose to this tool.
Do I get bold and italic back?: No. Inline emphasis inside a paragraph is not reconstructed. Headings and lists are, along with code blocks and paragraph breaks, and those are the parts carrying the document's structure.