The short answer: for a simple, born-digital PDF, the free pdf2md.morethan.io is perfectly adequate. For technical documents — formulas, tables, multi-column layouts, scanned pages — the gap between converters is enormous, and vision-model-based tools (KolmoPDF, Mathpix) are the only ones that consistently survive. Here's how all seven compare on the same three test documents.

How we tested

Three documents that cover the difficulty spectrum:

A two-column academic paper with display math and a results table.
A scanned book chapter (image-only, no text layer).
A financial report with merged-cell tables spanning page breaks.

Each converter was judged on reading order, table fidelity, formula handling, and how much cleanup the Markdown needed afterward.

The comparison at a glance

Converter	Type	Free tier	Two-column paper	Scanned pages	Complex tables	Math
KolmoPDF	VLM, web + API	Free credits	✅	✅	✅	✅ LaTeX
Mathpix	VLM, web + API	Limited	✅	✅	⚠️	✅ LaTeX
Marker	Open source	Fully free	✅	✅ slow	⚠️	⚠️
pdf2md.morethan.io	Text extraction	Fully free	⚠️	❌	❌	❌
MarkItDown	Open source	Fully free	⚠️	❌	❌	❌
CloudConvert	Cloud converter	25 conversions/day	⚠️	❌	⚠️	❌
iLovePDF	Cloud converter	Limited	⚠️	❌	⚠️	❌

1. KolmoPDF — best for technical documents

KolmoPDF parses pages with a Visual Language Model trained on papers, textbooks, and engineering docs. On our tests it was the only converter that handled all three documents without manual cleanup: the two-column paper came back in correct reading order with display math as $$ LaTeX, the scanned chapter parsed identically to a native PDF, and the merged-cell financial tables survived as pipe tables.

Pricing: 2 credits/page (3 with inline translation into 8 languages); free credits on signup.
Also has: a developer API with the same parser, and export back out via Markdown to PDF/Word conversion.
Weak spots: not offline, and per-page pricing beyond the free tier — for trivial PDFs, free tools do the job.

2. Mathpix — best for pure math workflows

Mathpix built its reputation on equation OCR, and it remains excellent: formula recognition is first-class and its Markdown (.mmd) ecosystem is mature. Table handling on our financial report was weaker, and the free tier is tight. If your documents are math-dense and tables are rare, it's a strong pick.

3. Marker — best open source

Marker chains layout detection, OCR, and equation models locally. Quality is genuinely good — clearly above every text-extraction tool — at the cost of heavy dependencies and real GPU time (the scanned chapter took minutes on CPU). For batch jobs on your own hardware with no per-page cost, it's the one to run. See our Python conversion guide for setup.

4. pdf2md.morethan.io — best free instant tool

Paste a PDF, get Markdown, no signup. For clean single-column digital PDFs it's all you need. It reads the text layer only: our two-column paper interleaved columns, the scan produced nothing, and tables flattened into run-on text. Know its lane and it's great.

5. MarkItDown — best for mixed-format pipelines

Microsoft's open-source everything-to-Markdown library. Its strength is uniformity (Office files, images, HTML, PDF through one interface), not PDF depth — expect text-extraction-grade results on hard documents.

6–7. CloudConvert & iLovePDF — fine for occasional simple files

Both are general-purpose conversion platforms with PDF→MD among hundreds of routes. Convenient if you already use them; neither is built for structural fidelity, and both stumbled on reading order and math. iLovePDF's strength remains its editing suite (merge, split, compress) rather than Markdown quality.

Which one should you pick?

Simple digital PDF, right now, free: pdf2md.morethan.io.
Technical documents, scans, tables, formulas: KolmoPDF — try the web converter with free credits.
Math-heavy, table-light: Mathpix.
Batch conversion, own hardware, $0: Marker.
One converter inside a many-format pipeline: MarkItDown.

FAQ

What's the best free PDF to Markdown converter? Fully free with no limits: pdf2md.morethan.io for easy files, Marker for hard ones if you can run Python. KolmoPDF's signup credits also convert typical documents free.

Why do most converters fail on two-column PDFs? They read the text layer in storage order, which rarely matches visual reading order. Vision-based parsers look at the rendered page instead, so columns come back in the sequence a human would read.

Can any converter handle scanned PDFs? Only those with OCR built in: KolmoPDF, Mathpix, and Marker in this list. Text-extraction tools see an empty text layer and return nothing.

Which output works best for Obsidian, Notion, or RAG? GitHub-Flavored Markdown with real heading structure and pipe tables — headings drive both Obsidian outlines and RAG chunk boundaries. That's the quality bar to test any converter against.

Best PDF to Markdown Converters in 2026 (7 Tested)