Research paper
Multi-column layout, figures, and citations.
Original source · arXivDrop a PDF here
Or choose a file. The browser analyzes it first; conversion sends the PDF to Cloudflare.
Markdown appears here once a PDF is loaded.
No pages sent yet · Local analysis
Why this converter
A PDF to Markdown converter that shows its sources
This online PDF to Markdown converter uses PDF.js on your device to inspect text layers, pages, and source geometry. Paid cloud conversion then sends text PDFs to Cloudflare's document converter to produce Markdown you can copy, download, or send into Obsidian, MkDocs, GitHub, and RAG workflows.
Unlike a typical PDF to Markdown converter, it keeps the provider method and Markdown offsets for every block. Cloudflare page markers add page-level provenance; any confidence Moondream returns is retained. Coordinates and confidence are shown only when a source actually provides them, never invented.
Simple digital PDFs — papers, manuals, contracts, exports from Word or LaTeX — follow the Cloudflare text-conversion path. Image-only PDFs are rendered into page images in your browser and sent to Moondream 3.1 for OCR. Word .docx conversion remains entirely local.
Local analysis · Cloudflare conversion
- Local
- PDF routing and source preview
- Cloudflare
- Text PDF to Markdown conversion
- Moondream 3.1
- OCR for scanned PDF pages
Prove it works on real documents
Each card loads an actual PDF through the same routing and conversion path used for your files.
Financial statement
Audited statements, dense tables, and footnotes.
Original source · SEC EDGARChinese technical report
Chinese headings, numbered sections, and punctuation.
Original source · MOST ChinaA better conversion loop
Fast by default. Precise when needed.
The browser inspects up to the first 30 pages to choose a route: text PDFs go to Cloudflare conversion, while fully image-only PDFs go to Moondream 3.1 OCR.
1
Inspect locally
Detect text layers, reading order, and scanned pages in your browser.
2
Route to the cloud
Send text PDFs to Cloudflare conversion or a fully scanned PDF's rendered pages to Moondream OCR.
3
Verify the output
Click a Markdown block to view its page when the provider returned one; otherwise it is marked as document-level.
Cloud PDF conversion and OCR require page points. Browser-local Word conversion remains public and costs zero points. See page points and Pro pricing
Markdown that fits where it is going
Run the PDF to Markdown converter once, then export clean Markdown or a profile that preserves the metadata your workflow needs.
Guides
Learn from real inputs and outputs.
Each PDF to Markdown converter guide includes a downloadable sample, expected output, failure modes, and fixes — not generic keyword filler.
Keep tables intact in Markdown
When to use GFM tables, HTML, or source-map JSON.
Read guideRoute scanned PDFs through OCR
Detect missing text layers, review reading order, and understand point costs.
Read guidePrepare PDFs for Obsidian
Frontmatter, a table of contents, and block ids that link back to the page.
Read guideFAQ
PDF to Markdown questions
Common questions about converting PDF files to Markdown, privacy, quality, and limits.
How do I convert a PDF to Markdown?
Sign in with Google, then choose a PDF at the top of this page or open the built-in sample. The browser analyzes its pages first. A text PDF is sent to Cloudflare for Markdown conversion; a scanned PDF is rendered into page images and sent to Moondream 3.1 OCR. The result appears next to the original page for review and export.
Is this PDF to Markdown converter free?
Browser-local Word conversion is free. Cloud PDF conversion, scanned-page OCR, and image recognition require sign-in plus a positive point balance or an active Pro plan. Results are not saved automatically, so copy or download them before leaving.
Do my files get uploaded to a server?
Yes for PDF conversion. Page analysis and preview happen in your browser, then text PDFs are uploaded to Cloudflare's document converter. Scanned PDFs are rendered locally and the resulting page images are sent to Moondream 3.1 OCR. Word .docx conversion remains on your device.
Can it convert scanned PDFs and images?
Yes. A scanned PDF has no usable text layer, so pdfmd renders its pages as images in the browser and sends them to Moondream 3.1 OCR instead of returning empty Markdown. Single photos use the same cloud recognition model through the image converter.
Are tables and formulas preserved?
Simple tables may convert to GitHub-flavored Markdown. Merged cells, multi-row headers, formulas, and tables spanning pages can be flattened or misread, so compare important output with the original PDF. The source-map export preserves only the page markers, offsets, and other metadata the converter actually returned; it does not diagnose merged cells.
What can I do with the Markdown afterwards?
Paste it into Obsidian, Notion, or any Markdown editor, commit it to a docs site such as MkDocs or Docusaurus, or chunk it for retrieval-augmented generation. The Obsidian export adds frontmatter, a table of contents, and anchors back to the original page.
Privacy you can see.
The browser analyzes PDFs for routing and preview; conversion then sends text PDFs to Cloudflare or a fully scanned PDF's rendered pages to Moondream 3.1.
Read how PDF routing and scanned-PDF OCR workRead the privacy policy