MarkItDown

Open source alternative to Unstructured.io, Google Document AI and Amazon Textract

191.3K stars
MIT
Active this month
Last commit 18 days ago

Last verified Oct 20, 2026

MarkItDown is a lightweight Python utility for converting PDFs, Word docs, Excel files, images, audio, and more to Markdown for LLM pipelines.

Visit website GitHub191.3K

Read the full breakdownWho it is for, how it works, trade-offs, the quick start and the FAQ

who it's for

Who MarkItDown is for#

AI engineers building RAG ingestion

MarkItDown converts common document formats into Markdown before chunking and embedding.

Skip if:

your source data already arrives as clean Markdown or JSON.

Data teams normalizing office documents

the CLI can batch convert Word, Excel, PowerPoint, and PDF files.

Skip if:

you need exact visual layout preservation.

Developers adding document upload to an AI app

the Python API fits backend conversion steps.

Skip if:

users upload untrusted files and you cannot sandbox conversion.

the problem

The problem it solves#

LLM pipelines break when source documents arrive as PDFs, slide decks, spreadsheets, images, audio files, or mixed archives. Raw extraction often loses headings, tables, links, and other structure that helps models understand the content.

Teams usually patch this with one parser per file type, then spend time normalizing output before indexing or analysis. That slows down ingestion work and creates security concerns when untrusted files run through broad conversion code with too many permissions.

how MarkItDown solves it

How it solves it#

Broad file conversion

Converts PDF, PowerPoint, Word, Excel, images, audio, HTML, CSV, JSON, XML, ZIP archives, YouTube URLs, and EPUBs into Markdown.

CLI pipeline support

CLI usage supports shell pipelines, such as converting a PDF and redirecting Markdown output to a file.

Python API

Python API exposes MarkItDown conversion inside ingestion jobs, evaluation scripts, and document-processing services.

LLM-ready Markdown output

Markdown output preserves useful structure such as headings, lists, tables, and links for LLM and search workflows.

Optional PyPI extras

PyPI package supports optional extras through `markitdown[all]` for broad file-format coverage.

strengths · trade-offs

Strengths and trade-offs#

Strengths

Built for LLM ingestionMarkItDown focuses on LLM-ready Markdown rather than pixel-perfect document reproduction, which matches RAG and text analysis needs.
Messy input coverageThe supported format list covers the messy inputs teams actually receive, including Office files, images, audio, archives, and web formats.
MIT licenseThe MIT license makes it practical to embed in internal ingestion systems and commercial AI products.
CLI and API workflowsThe CLI and Python API support both quick local conversions and repeatable batch pipelines.

Trade-offs

-Not for pixel-perfect conversionMarkItDown is not meant for high-fidelity human-facing document conversion, so layout-heavy PDFs or slide decks may need a different tool.
-Untrusted inputs need sandboxingThe project warns that conversion runs with the privileges of the current process, so untrusted inputs require sandboxing and narrow conversion functions.
-Optional dependencies requiredOCR, audio transcription, and optional format support depend on installed extras and external dependencies.

install · quick start

Quick start#

bash

pip install markitdown[all]
markitdown path-to-file.pdf > document.md

from markitdown import MarkItDown

md = MarkItDown()
result = md.convert("test.xlsx")
print(result.text_content)

tech stack · detected from GitHub

What it's built on#

Languages: Python
Frameworks: LangChain

frequently asked

FAQ#

What file types does MarkItDown support?

MarkItDown supports PDFs, Word, PowerPoint, Excel, images with OCR metadata, audio with transcription metadata, HTML, CSV, JSON, XML, ZIP files, YouTube URLs, EPUBs, and more. Exact support depends on installed extras and optional dependencies.

Is MarkItDown good for human-readable document conversion?

Not as the primary goal. MarkItDown optimizes for Markdown that works well in LLM and text analysis pipelines, not perfect visual reproduction for people.

Is MarkItDown safe for untrusted files?

Use caution. The project states that MarkItDown performs I/O with the privileges of the current process, so untrusted inputs should run in a restricted environment with narrow conversion functions.

also worth a look

Similar open-source tools#

PaddleOCR

Alternative toGoogle Document AI

Self-hosted OCR toolkit turning PDFs and images into LLM-ready data

91,289 GitHub stars
Licence Apache-2.0

pdf-inspector

Alternative toAdobe PDF Extract API

Rust-powered PDF parser with smart OCR routing for AI pipelines

19,698 GitHub stars
Licence MIT

Marker

Alternative toUnstructured.io

Convert documents to clean markdown and JSON with open source AI.

40,566 GitHub stars
Licence Apache-2.0

Unstructured

Alternative toLlamaParse

Document parsing and ETL for LLMs, RAG pipelines, and AI agents.

15,583 GitHub stars
Licence Apache-2.0

Hermes Agent

Featured

Alternative toGrok Bot

Self-hosted AI agent: persistent memory, multi-channel chat, model choice.

255,514 GitHub stars
Licence MIT

Vllm

Alternative toTogether AI

High-throughput LLM inference server with OpenAI-compatible API

94,441 GitHub stars
Licence Apache-2.0

by category

Open source tools in these categories#

404

Featured

rakazo

Alternative toGrok Bot

AI teammates you own: your keys, your model, your machine.

4,904 GitHub stars
Licence Apache-2.0
#AI Agents

Omnara

Alternative toClaude Managed Agents

Open-source agent deployment API. Self-host or use Omnara Cloud.

2,957 GitHub stars
Licence Apache-2.0
#AI Agents

Inbox Zero

Alternative toSuperhuman

AI email assistant that drafts replies and clears your inbox

12,712 GitHub stars
Licence AGPL-3.0
#Email

Feature your project

The projects we add and every new guide. Free, unsubscribe any time.

Find open source alternatives to paid software

Who MarkItDown is for#