Overview: Convert PDF to clean Markdown for ChatGPT, Claude and RAG pipelines. Headings, lists, tables and page markers kept. Runs in your browser, no upload.
PDF to Markdown for LLM and RAG
Turn a text-based PDF into clean Markdown for ChatGPT, Claude or your vector database. Conversion runs in your browser, so your file is never uploaded.
Drop a PDF here or click to choose one
Text-based PDFs only. Scanned pages need OCR first.
What it keeps
The converter reads the PDF's text layer and rebuilds structure from layout: headings from font size and font weight relative to body text, wrapped lines merged into real paragraphs, hyphenated line breaks repaired, bullet and numbered lists, preserved hyperlinks, and tables that line up in clear columns (with a proper header row). Page markers help you trace a retrieved chunk back to its source page.
How it compares
| PDFTeq | Docling, Marker, MarkItDown | |
|---|---|---|
| Setup | None, open the page | Python install, models to download |
| Privacy | File stays in your browser | Local too, once installed |
| Scanned PDFs / OCR | No | Yes |
| Complex layouts, equations | Basic | Much better |
If you process scanned documents or dense academic layouts at scale, use one of the Python tools. For quick conversions of text-based PDFs, this is faster.
Related resources
- Guide: how to convert a PDF to Markdown for LLMs and RAG
- Read PDF Aloud, which includes OCR for scanned pages.
- Split PDF to convert only the chapters you need.
- Compress PDF before sharing the original.
Frequently asked questions
Is my PDF uploaded anywhere?
No. The file is read and converted by JavaScript in your browser tab. Nothing is sent to a server.
Can it convert scanned PDFs?
No. This converter reads the PDF's text layer, so scanned pages come out empty. Run OCR first, or use our Read PDF Aloud tool, which includes in-browser OCR.
Does it support two-column research papers?
Yes. Two-column pages are automatically detected and read column by column (left column first, then right column) when a clear vertical gutter exists. Highly complex, irregular or mixed layouts can still sometimes mis-order.
How does it compare with Docling, Marker or MarkItDown?
Those Python tools handle complex layouts and OCR better, but need installation. This tool needs nothing and keeps files on your device, which suits quick conversions of text-based PDFs.
How are tokens estimated?
As roughly four characters per token. It is a quick guide, not an exact count for any specific model.
Does it keep tables, links and images?
Tables that line up in clear columns keep a proper header row. Hyperlinks are preserved as clean Markdown links when the PDF has link annotations (supporting http, https, and mailto). Images and equations are not preserved.