MarkItDown

Convert common documents into Markdown

MarkItDown converts documents and other source material into Markdown for editing, search, publishing or use in AI workflows. Supported inputs include PDF; Word DOCX; Excel XLS and XLSX; PowerPoint PPTX; HTML and web pages; CSV and TSV; JSON and XML; plain text and Markdown; EPUB; and Jupyter notebooks. Image formats such as PNG, JPG and WebP can also be processed with OCR. The resulting Markdown can preserve useful structure such as headings, tables, links, code and mathematical notation.

Review and refine the output

The workflow provides a preview of the converted text so users can inspect the result before exporting a .md file. This is useful when moving material into a documentation site, knowledge base, note-taking tool or text-based publishing workflow. Since conversion quality depends on the layout and source file, tables, scans and complex formatting should be checked before reusing the output.

Local processing, server conversion and OCR

• Browser-local conversion: Supported formats can be processed on the device, keeping those files in the browser rather than uploading them.
• Server conversion: EPUB files, Jupyter notebooks and scanned documents use server processing. These uploads are held temporarily and then deleted, according to the service’s privacy description.
• OCR choices: Local OCR is available for images, while an AI OCR option sends the image to an OCR service for temporary processing. Choose the local path when keeping image data on the device is a priority.

The service offers everyday conversion without an account. Signing in unlocks saved results, batch processing, larger files and higher-capacity OCR; optional Pro features are described as paid server and AI OCR usage. Check current file-size, batch and privacy terms before sending sensitive material.

Independent project, not a Microsoft product

MarkItDown is presented as an independent service built around Microsoft’s open-source MarkItDown library. It is not an official Microsoft site and does not claim Microsoft endorsement. The practical choice is between quick, local conversion for supported formats and server or AI-assisted processing for formats or scans that need those additional capabilities.