Docling vs Zerox
Which should you use? A side-by-side look at what each one does, who it suits, what it costs, how hard it is to set up, and where each falls short.
Open source · MIT
Python library that converts PDFs and office files into structured text for AI use.
Open source · MIT
Converts documents to Markdown by sending page images to vision AI models.
The short answer
Choose Docling if you need…
- Python developers building retrieval or RAG pipelines over PDFs and office files
- Teams that need tables and reading order preserved when extracting document content
- Users who need local processing for sensitive or air-gapped documents
- People who want a command-line tool to convert a document to Markdown
Choose Zerox if you need…
- Developers building pipelines that feed documents into AI or search systems
- Teams with PDFs containing tables, charts or unusual layouts
- Users who already have API access to OpenAI, Azure OpenAI, AWS Bedrock or Google Gemini models
- Projects that need structured data pulled from documents using a JSON Schema
Side-by-side
| Docling | Zerox | |
|---|---|---|
| Type | Library | Library |
| Price | Free and open source | Free and open source |
| Licence | MIT | MIT |
| GitHub stars | 70k | 12.3k |
| Setup difficulty | Beginner | Intermediate |
| How you get it | Install with a package manager | Install with a package manager |
| Main language | Python | TypeScript |
| Last code update | 5 days ago | 18 months ago |
What each one is
Docling
Docling parses many document formats, including PDF, Word, PowerPoint, Excel, HTML and email, and turns them into a structured representation. It exports that content to formats such as Markdown, HTML and JSON. People use it to prepare documents for generative AI and retrieval pipelines, and it can run locally for sensitive data.
Zerox
Zerox is a Node.js and Python package that turns documents such as PDFs, DOCX files and images into Markdown. It converts each page into an image, sends each image to a vision model like GPT-4o and asks for Markdown, then joins the results. The README says this approach handles unusual layouts, tables and charts better than traditional text extraction, which makes it useful for preparing documents for AI systems.
Best for
Docling
- Python developers building retrieval or RAG pipelines over PDFs and office files
- Teams that need tables and reading order preserved when extracting document content
- Users who need local processing for sensitive or air-gapped documents
- People who want a command-line tool to convert a document to Markdown
Zerox
- Developers building pipelines that feed documents into AI or search systems
- Teams with PDFs containing tables, charts or unusual layouts
- Users who already have API access to OpenAI, Azure OpenAI, AWS Bedrock or Google Gemini models
- Projects that need structured data pulled from documents using a JSON Schema
Strengths
Docling
- Supports a wide range of input formats in one library, per the README feature list
- Can run locally, which the README highlights for sensitive data and air-gapped setups
- Includes a CLI, a Python API, an MCP server and an API server option
Zerox
- Simple, short API for turning a file into Markdown
- Supports several model providers, including OpenAI, Azure OpenAI, AWS Bedrock and Google Gemini
- Offers options for structured extraction, per-page processing, concurrency and error handling
Not ideal for
Docling
- Users who need a no-code desktop app, since the README centers on pip install and CLI or Python use
- Projects that need metadata extraction such as title and authors, which the README lists as coming soon
- Chemistry-heavy work, since molecular structure understanding is also listed as coming soon
Zerox
- People who need a no-code tool: the README describes an SDK used in code, and the hosted demo is the only non-coding option mentioned
- Users who want to avoid per-page model API costs, since every page is sent to a paid vision model
- High-volume or offline processing, because the README does not describe local or offline models
Trade-offs
Docling
- Advanced features such as VLM pipelines and audio or video parsing may need extra setup or models; the README does not detail requirements for each
- Some features are newer, as shown in the What's new section, so behavior may change between releases
Zerox
- Requires external system tools (GraphicsMagick and Ghostscript for Node, Poppler for Python) and a paid model API key
- Sends document page images to third-party model providers, so privacy depends on that provider
- The README is truncated, so some model lists and details could not be fully verified
Getting started
Zerox
Install the Node package
npm install zeroxInstall GraphicsMagick on Linux
sudo apt-get install -y graphicsmagickFrequently asked questions
What is the difference between Docling and Zerox?
Docling: Python library that converts PDFs and office files into structured text for AI use. Zerox: Converts documents to Markdown by sending page images to vision AI models. Docling suits python developers building retrieval or rag pipelines over pdfs and office files, while Zerox suits developers building pipelines that feed documents into ai or search systems.
Are Docling and Zerox free?
Docling: Free and open source. Zerox: Free and open source.
Which is easier to set up, Docling or Zerox?
Docling is easier to get started with (rated beginner); Zerox is rated intermediate and usually needs more setup.
Is Docling or Zerox more popular?
Docling has more GitHub stars (70k) than Zerox (12.3k). Popularity isn't the same as fit — check which one matches your use case above.
What are other alternatives to Docling and Zerox?
Other options people compare them with include MarkItDown, pdf-inspector, Rich, Showdown, remark.
Keep comparing
Built from our catalogue entries for both products: open-source details come from each project's README (AI-assisted, refreshed weekly) and GitHub data; AI-tool details are editorial. Spotted something wrong? Let us know from either product's page.