Inspect and estimate
Markovo checks the page count and input boundary before execution. You see the estimate first, and a confirmed job cannot silently spend beyond its approved maximum.
PDF to Markdown Converter
Your PDFs. Ready to work.
Page-aware Markdown, ready for your next step.
Delivery ready
Your Markdown is ready
Verified output links appear here after conversion.
Optional: download audit files for verification or troubleshooting.
Delivery ready
Your link, in Markdown
Conversion details & limits
PDF to Markdown Converter for Clean, Reviewable Output.
Convert PDF to Markdown (.md) while preserving useful headings, paragraphs, common tables, links, and images in a page-aware bundle with source maps and quality evidence.
Preview a PDF up to 5 estimated Credits without an account. Signed-in jobs share the same Credits and History across Web, API, CLI, and MCP.
01 / How it works
How PDF to Markdown conversion works.
The converter uses a bounded job flow instead of returning an untraceable block of pasted text. It inspects the input, estimates the maximum Credit cost, selects the current PDF route, and packages the result with files that help you review where the Markdown came from.
Convert page-aware content
Recoverable headings, paragraphs, links, common tables, and useful figures become Markdown and assets. Page references are retained where the route can support them.
Review the bundle
Open the Markdown preview, download the verified ZIP, inspect warnings, and compare uncertain content with the source PDF before using it in production.
02 / What you get
What the PDF Markdown bundle preserves.
A useful conversion needs more than readable prose. Markovo keeps the main Markdown beside supporting evidence so editors, developers, and AI workflows can distinguish extracted content from review signals.
Structured Markdown
output.md carries the recovered reading order, headings, paragraphs, links, lists, and common tables. It is suitable for editing, search preparation, or downstream chunking after review.
Sources and assets
source_map.json connects recoverable output to source pages. Reliably extracted figures live under assets/ and use relative links so they remain portable with the bundle.
Quality evidence
quality_report.json, metadata, and the checksummed manifest record warnings and delivered files. They do not claim perfection; they make review boundaries visible.
03 / Source details
Digital PDFs, scans, tables, and formulas have different boundaries.
PDF is a page description format, so two files that look similar can expose very different underlying structure. The correct expectation depends on whether text, images, reading order, and table geometry can be recovered reliably.
Available
Digital PDFs with selectable text
Ordinary reports, manuals, and papers with selectable text are the strongest current route. Headings, paragraphs, links, lists, and straightforward page order are usually the most reviewable output. Multi-column reading order, footnotes, and floating captions still deserve comparison with the source.
Review carefully
Common tables and extracted figures
Recoverable tables can become structured Markdown rather than screenshots. Reliably exposed figures can be stored as assets and linked from output.md. Merged cells, nested headers, charts, diagrams, and visually positioned labels may not preserve their full meaning.
OCR lane
Scanned text and formulas
Image-heavy PDFs and formula-rich pages go through OCR in the standard pipeline. Low-resolution scans, handwriting, unusual symbols, dense equations, and complex layouts can still be wrong. Markovo reports weak output visibly and requires review against the original.
04 / Connect your tools
Use the PDF to Markdown Converter via Web, API, CLI, or MCP.
Every client uses the same pdf-to-markdown capability, estimate-first job contract, ownership checks, Credit balance, History, and bundle format. The browser supports the guest trial; API, CLI, and MCP conversion use an account and API key.
REST API
curl -X POST https://markovo.net/v1/convert \
-H "Authorization: Bearer ${MARKOVO_API_KEY}" \
-F "file=@paper.pdf" \
-F "capability_id=pdf-to-markdown" \
-F "max_credits=30"Request an estimate, confirm the bounded maximum, then poll the returned job with its owner-authorized credential.
CLI
markovo convert paper.pdf \
--out runs/paper \
--max-credits 30The CLI downloads and verifies the same bundle rather than inventing a separate local output format.
MCP
markovo_convert({
"input_path": "paper.pdf",
"max_credits": 30
})MCP exposes the same capability and limits to an authorized agent. Review remains a user responsibility when the report flags uncertain content.
05 / Ways to use your Markdown
Prepare PDF Markdown for AI, RAG, and editing.
Clean Markdown is easier to diff, search, chunk, and edit than opaque PDF extraction, but conversion is only one stage of a reliable workflow. Keep the bundle intact, inspect warnings, and decide which evidence your downstream use requires.
Grounded AI input
Use headings and page-aware mappings to build reviewable context for ChatGPT, Claude, or an internal assistant. Do not treat a converted formula, chart, or OCR passage as verified solely because it is valid Markdown.
Search and chunking
Chunk after checking section order, table continuity, and repeated headers. Preserve source mappings or page metadata when retrieval results need to point a reviewer back to the original PDF.
Publishing and migration
Edit output.md while keeping assets beside it. For compliance, legal, scientific, or financial material, compare the final text and numbers with the original before publication.
A 60-second review order: open quality_report.json, jump to its lowest-confidence pages, then compare names, totals, and dates against the source.
PDF to Markdown: what converts cleanly, what needs review.
Every page gets a confidence score in quality_report.json — low-confidence pages are flagged for your review instead of silently approximated. These are the PDF-specific cases that decide the score.
Scanned pages
Scans route through OCR. Handwriting and low-DPI pages land in the review queue with per-page confidence scores.
Multi-column layouts
Reading order can merge columns on dense magazine layouts — check ordering against the source.
Equations & formulas
Formulas go through OCR processing like scans. Complex notation should be checked against the source.
Charts
Preserved as image assets where recoverable — the underlying data is not re-digitized.
Password-protected PDFs
Encrypted PDFs are rejected during validation — remove the password in your PDF tool first. Markovo does not attempt to open protected files or approximate their contents.
Mixed digital & scanned pages
One PDF can contain both text-layer and scanned pages. Per-page routing merges them into a single output.md with per-page provenance.
Markovo routes per page, not per file: digital text pages take the fast text-layer route, scanned and layout-heavy pages take the document-model OCR route — you get one merged output.md with per-page provenance. Format references: CommonMark · Adobe PDF Services API docs. Working with scans? See Image to Markdown and the OvisOCR2-class pipeline.
PDF to Markdown questions.
Current production behavior and explicit quality boundaries.
What is a PDF to MD converter?
A PDF to MD converter turns a PDF document into a Markdown (.md) file. Markovo keeps recoverable headings, paragraphs, links, common tables, and images in a reviewable bundle, with source maps and quality evidence where available.
What kinds of PDFs work best with the PDF to Markdown converter?
Digital PDFs with selectable text, ordinary headings, paragraphs, links, and common tables are the strongest current route. Scans, handwriting, formulas, and dense visual layouts go through OCR processing and require manual review.
Is the PDF to Markdown converter free?
Guests can convert and review two PDF queues per UTC day before signing in, up to 5 estimated Credits. Each guest PDF is limited to 25 MB and 50 pages. Signed-in Free accounts have a separate 25-page limit. Download requires an account and settles completed work once against account Credits. A free account includes 25 monthly Credits, one API key, and up to five daily files.
Does PDF to Markdown preserve tables and images?
Markovo preserves common tables as structured Markdown when the source is recoverable and stores reliably extracted figures as linked assets. Complex tables, charts, and unusual reading order still require comparison with the source PDF.
Can Markovo convert scanned PDFs and formulas?
OCR handles scans and formulas through the standard pipeline. It can recover useful content, but weak scans, handwriting, dense equations, and complex layouts can produce warnings and always require review against the original PDF.
Can I use PDF to Markdown through an API, CLI, or MCP server?
Yes. Web, REST API, CLI, and MCP use the same capability ID, estimate, Credit balance, job history, ownership checks, and downloadable bundle. API, CLI, and MCP conversion require an account and API key.
What files are included in a PDF Markdown bundle?
A successful bundle can include output.md, extracted assets, metadata, source_map.json, quality_report.json, and a checksummed manifest. The exact contents depend on what the source PDF exposes and which processing route runs.
Can I convert a password-protected or encrypted PDF?
No. Encrypted PDFs are rejected during validation instead of being partially converted. Remove the password locally, then upload the unprotected file.
Why does the estimate differ between two PDFs of similar size?
The estimate reflects what the file actually contains — page count and how many pages need the OCR route instead of the text-layer route — not just file size. A confirmed job cannot spend beyond its approved maximum.
