PDF a Markdown
Convierte documentos PDF en formato Markdown estructurado.
Click to select or drop a PDF here
Processed entirely in your browser — never uploaded
Most people want Markdown for what comes next: feeding a document to an AI as context, pasting it into Obsidian, or checking it into a docs repo where diffs matter. Copying straight out of a PDF flattens headings and body text into one blob, so the structure has to be reconstructed — font size, weight, and position decide what reads as a heading and what reads as a paragraph. That's Heuristic mode: plain rules, no network. Switch to AI Enhanced and I hand the text already extracted in your browser to DeepSeek to sort the hierarchy out; text goes out, the PDF itself doesn't.
The limit is 200MB, the same for free and Pro. Not generosity — this tool barely touches memory: pdf.js reads pages on demand, so peak usage stays near the file's own size. The trade-off is that it only understands text. Tables come apart, formulas don't become LaTeX, images aren't extracted. A scan with no OCR layer converts to an empty file; run it through OCR first.
FAQ
Is there a file size limit when converting a PDF to Markdown?
200MB, and it's the same number on free and Pro — no tiering here. Some tools on this site do raise the cap for Pro, because they load and rewrite the entire PDF in memory, where the ceiling is real. This one doesn't: text extraction reads pages on demand, so peak memory stays close to the file's own size. Page count hardly matters either — a 600-page manual is just slower.
Does converting a PDF to Markdown upload my file to a server?
In Heuristic mode, no. pdf.js reads the file, builds the Markdown, and hands you the download inside the tab; it works with your Wi-Fi off, and you can verify that yourself the same way as with the other tools on this site. AI Enhanced is different: the text already extracted locally is sent to DeepSeek (the UI says "DeepSeek · 5 free per day"). Text leaves, the PDF never does. For a contract you'd rather keep sealed, stay on Heuristic.
Can I convert a scanned PDF to Markdown?
It depends on whether the scan has a text layer. Plenty of scans have already been through OCR in a scanner or in Acrobat, and those convert fine — heading detection just tends to be rougher. A scan with no OCR layer is only images as far as pdf.js is concerned, so you get an empty file. Run it through the OCR tool first, then come back here.
How does PDF to Markdown conversion handle tables, formulas, and images?
Badly, mostly — it's the weakest part of this tool. A table in a PDF is just text placed at coordinates; there are no rows or columns to read back. Clean ruled tables sometimes survive; merged cells, borderless layouts, and tables spanning a page break usually collapse. Formulas come out as scattered characters, never LaTeX. Images aren't extracted at all. If table fidelity matters, plan on fixing it by hand.
What's the difference between Heuristic and AI Enhanced mode in PDF to Markdown?
Start with Heuristic: it goes by layout rules, runs offline, and costs nothing per page. For tidy reports, papers, and manuals it's usually enough. If headings and body text are set at nearly the same size, or the hierarchy comes out scrambled, try AI Enhanced — DeepSeek re-reads the extracted text and rebuilds the structure, 5 free runs a day. It isn't guaranteed to be better, just a different guess.