Convert a PDF to Markdown for ChatGPT, Claude, or RAG Context
Markdown is a cheaper, cleaner context format than a raw PDF dump. Convert first, strip the running headers, then paste the text into the model or your retrieval pipeline.
Direct answer
Convert the PDF to Markdown, then paste the Markdown into ChatGPT or Claude instead of attaching the PDF. Markdown carries the heading structure the model uses to navigate a document, and it drops the repeated page furniture that wastes tokens. Keep the 'Strip headers, footers, page numbers' toggle on so the same title bar is not repeated once per page in your context window.
Open PDF to Markdown converterWhen to use this
- You want a model to summarise or answer questions about a long document.
- You are chunking documents for a retrieval pipeline and need clean text with headings.
- The PDF attachment is too large, or the model handles the attached layout poorly.
Steps
- Convert the PDF in the browser converter.
- Leave 'Strip headers, footers, page numbers' on to remove the repeated title bar and page numbers.
- Leave 'Merge hyphenated words' on so words split across line breaks do not become two broken tokens.
- Scan the output for interleaved columns if the source is a two-column paper, and fix the order by hand.
- Copy the Markdown into the chat, or download the .md file for your chunking script.
Example
A 40-page PDF report where every page repeats the header 'Acme Confidential' and a page number in the footer.
Markdown with the repeated header and the page numbers removed, headings as # and ##, and paragraphs rejoined into full sentences instead of one line per printed line.
Common mistakes
- Do not feed a scanned PDF to this workflow. It has no text layer, so you would paste an empty document into the model.
- Do not skip a read-through of a two-column paper; interleaved columns confuse a model exactly as much as they confuse a human.
- Do not expect the tool to chunk, embed, or summarise. It only converts text to Markdown.
- Check that a table you rely on survived. Tables come back as lines of text, not pipe tables.
FAQ
- Why is Markdown better context than a PDF?
- Headings become explicit # markers the model can use for structure, and paragraphs are rejoined into real sentences. A raw PDF dump often arrives as one line per printed line with the header repeated on every page, which spends tokens on furniture rather than content.
- Does this reduce token count?
- Usually yes, because stripping repeated headers and footers removes text that appears once per page. The amount saved depends on the document; a 40-page report with a two-line header saves far more than a 3-page memo.
- Is my document private if it contains client data?
- The conversion is private: the PDF is read in your browser and never uploaded. What you do afterwards is not — pasting the Markdown into a hosted AI model sends it to that provider.
Related PDF to Markdown guides
How to Convert a PDF to Markdown
Drop a PDF into the browser converter. It reads the text layer, infers headings from font size, rebuilds lists, and returns Markdown you can copy or download as a .md file.
Why a Scanned PDF Converts to Nothing (OCR Explained)
A scanned PDF stores pictures of pages, not characters. Text extraction returns nothing because there is nothing to extract. Run OCR first, then convert.
Keep Headings and Structure When Converting a PDF
The converter infers headings from relative font size: the largest size becomes #, the next becomes ##. Here is when that works, and what to do when a document defeats it.
Extracting Tables From a PDF to Markdown: What Actually Works
A PDF table is positioned text with no cell structure. This converter returns the table's text as lines, not as a Markdown pipe table. Here is the honest workaround.
Convert a PDF to Markdown for Obsidian and Other Notes Apps
Download the converted .md file straight into your vault. Headings become real outline levels, and the source PDF never leaves your machine.