PDF to Markdown Parser — Tables, Formulas & Code Preserved

Intelligently convert PDF to editable Markdown, with table handling and multi-lang translation

Upload PDF File

Drag & Drop files here, or

PDF only, Max 300MB, Max 800 pages

Settings

Points Deduction Info

  • • PDF Parsing：2 pts/pg
  • • PDF Parsing+MD Translation：3 pts/pg

How PDF to Markdown conversion works

KolmoPDF uses a Visual Language Model (VLM) to analyze PDF pages and output structured Markdown that preserves the semantic content of the document.

What the parser handles

  • Multi-column layouts: Pages with two or more columns are read in the correct logical order, not left-to-right across the physical page.
  • Mathematical formulas: Inline and display math is output as LaTeX within the Markdown, for example $E = mc^2$ or$$\int_0^\infty e^{-x^2} dx$$.
  • Tables: Tables can be output as Markdown tables or preserved as images. The experimental cross-page table merge feature detects and merges tables that span page breaks.
  • Code blocks: Source code and technical notation is wrapped in fenced code blocks with language hints where detectable.
  • Translation: Enable translation to convert document content into Chinese, Japanese, Korean, French, German, Spanish, or Russian during parsing. Adds 1 credit/page.

File limits

PDF files only. Maximum file size: 300 MB. Maximum page count: 800 pages.

What this PDF to Markdown converter does

KolmoPDF parses technical PDFs into structured Markdown using a Visual Language Model trained on academic papers, technical books, and engineering documents. It preserves multi-column reading order, recognizes LaTeX formulas inline and in display form, reconstructs tables that span page breaks, and keeps code blocks intact. Output is clean Markdown that flows into static-site generators, knowledge bases, and downstream translation pipelines.

Key capabilities

Common use cases

Researchers use it to extract LaTeX from papers for note-taking systems. Engineering teams pipe extracted Markdown into Confluence, Notion, or internal wikis. Localization teams feed the Markdown into translation memory tools. Educators rebuild slides and study materials from textbook PDFs without retyping equations. AI teams convert PDF libraries to Markdown as the ingestion step for RAG pipelines and LLM fine-tuning corpora.

How to convert a PDF to Markdown

Converting a PDF to Markdown takes under a minute:

  • Upload your PDF — native or scanned, up to 300 MB. No installation required; the converter runs entirely in your browser.
  • Choose options: keep the original language, or translate to English, Chinese, Japanese, Korean, French, German, Spanish, or Russian in the same pass.
  • Parsing runs page by page. Multi-column layouts, formulas, tables, and code blocks are detected automatically — nothing to configure.
  • Download the result as a .md file, or copy the Markdown straight into Obsidian, Notion, a Git repository, or your RAG ingestion pipeline.

Why convert PDF to Markdown for LLMs and RAG?

PDF is a layout format, not a content format — raw extracted text scrambles reading order, drops table structure, and turns equations into garbage characters. Markdown is the format LLMs handle best: heading hierarchy becomes chunking boundaries, tables stay machine-readable, and LaTeX formulas remain meaningful tokens. If you are preparing documents for Claude, ChatGPT, or a vector database, converting PDF to clean Markdown first measurably improves retrieval quality and answer accuracy. KolmoPDF also exposes this converter as a developer API for batch processing.

PDF to Markdown converter vs. copy-paste or Pandoc

Copy-pasting from a PDF loses structure entirely and mangles multi-column pages. Pandoc is excellent for many formats but does not perform OCR or layout analysis, so PDF input generally fails on scanned files, complex tables, and math. Rule-based extractors handle simple single-column text but break on the technical documents that matter — papers, manuals, textbooks. A Visual Language Model reads each page the way a human does, which is why KolmoPDF holds structure where rule-based converters drop it.

Frequently asked questions

Does it work on scanned PDFs?

Yes. The Visual Language Model treats text as image regions, so both native and scanned PDFs are parsed the same way — no separate OCR step needed.

How much does it cost per page?

PDF parsing is 2 credits per page; parsing with inline translation is 3 credits per page. New users receive 100 free credits on signup, so you can convert PDF to Markdown free before deciding to upgrade.

What is the maximum file size?

Free and paid plans accept files up to 300 MB. Page count limits depend on your remaining credit balance.

Does the converter keep images from the PDF?

Yes. Figures and images are extracted and referenced from the Markdown output, so converting a PDF to Markdown with images intact works out of the box.

Can I convert PDF to Markdown for Claude, ChatGPT, or a RAG pipeline?

That is one of the most common uses. Clean Markdown with real heading structure chunks better in vector databases and reads better as LLM context than raw PDF text extraction. For batch conversion, use the developer API.

How is this different from pdf2md or other free online converters?

Most free converters are rule-based text extractors: fine for simple one-column text, unreliable for multi-column layouts, merged-cell tables, LaTeX math, and scanned pages. KolmoPDF uses a Visual Language Model trained on technical documents, so exactly those hard cases are what it is built for.

Is there an API for PDF to Markdown conversion?

Yes. The same parser is available as a REST API with per-page pricing — see the API documentation for endpoints, code samples, and rate limits.

What Markdown flavor does the output use?

GitHub-Flavored Markdown (GFM): standard headings, pipe tables, fenced code blocks, and LaTeX math inside $ and $$ delimiters — compatible with Obsidian, Notion import, MkDocs, and most static-site generators.