DS4SD

docling: Parse Documents for Generative AI

Docling is a Python document-processing library that turns files such as PDFs, office documents, images, and audio into structured content for generative AI workflows. It supports local processing and runs on macOS, Linux, and Windows.

ActivePythonMITAILibraryPDF
Stars
69.6k
Forks
5.2k
Stars, last 30 days
+1,651
Last commit
today
Contributors
100+

Our take

Activeregular commits and releases

Health score

76/100

How it is scored
Activity
88
Community
92
Issues
68
Pull requests
54

+1,651 stars in the last 30 days

  • Releases more often than 93% of the projects we track
  • Merges more pull requests than 86% of the projects we track
  • More contributors than 78% of the projects we track

Docling is a capable, actively maintained document-ingestion library with broad format support and a clear fit for Python-based generative AI pipelines.

Good fit if

  • You need structured extraction from PDFs and other document formats rather than basic text conversion.
  • Your application uses Python and benefits from local processing or the provided AI-framework integrations.
  • You want an MIT-licensed codebase with frequent releases and contributions from a broad contributor base.

Look elsewhere if

  • You need a non-Python runtime, since Docling requires Python 3.10 or later.
  • You need a specific capability that the project lists as coming soon, such as metadata extraction for titles, authors, references, and language.
  • You require every component to share the codebase's MIT license, since individual model packages have their own licenses.
All health signals
Last commit2026-10-25 (today)
Commits, last 90 days100+
Releases, last 12 months99 (latest v2.142.1, 2026-10-23)
Contributors100+ (top contributor: 18% of commits)
Issues closed, last 90 days19+ (typically closed in 3 days)
Pull requests merged, last 90 days79 (typically merged in 2 days)
Project age2 years

Checked on 2026-10-26 with the GitHub API.

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

70,722

Forks

5,372

Change since Nov 17, 2026: +49 stars·+8 forks

Overview

Docling is a Python document-processing library for preparing diverse files for generative AI workflows. It parses documents, including PDFs with layout and table structure, and converts their content into a unified representation that can be exported in formats such as Markdown and JSON.

It is aimed at developers building document-aware AI applications who need more than plain text extraction. Docling runs on macOS, Linux, and Windows, and can process data locally, including in air-gapped environments.

Key Features

  • Parses PDFs, DOCX, PPTX, XLSX, HTML, EPUB, images, audio, video, email, and other formats.
  • Analyzes PDF layout, reading order, tables, code, formulas, and images.
  • Exports structured content to Markdown, HTML, WebVTT, DocTags, and lossless JSON.
  • Supports OCR for scanned PDFs and images.
  • Provides integrations with LangChain, LlamaIndex, Crew AI, and Haystack.
  • Offers a CLI, a Python API, an MCP server, and an API server.
  • Supports visual language models and automatic speech recognition models.

Use Cases

  • AI application developers can convert PDFs and office files into structured input for retrieval or question-answering workflows.
  • Teams handling scanned documents can use OCR and layout analysis to recover readable text and tables.
  • Researchers and analysts can extract content from formats such as scientific PDFs, EPUB books, or XBRL financial reports.
  • Developers building media workflows can process audio and video with transcripts and representative keyframes.

What you need

Detected in the repository

  • Python >=3.10,<4.0 (from pyproject.toml)
  • A Dockerfile, so it can run in a container
  • A test suite and automated checks on GitHub Actions

License in plain words

MITpermissive

  • Commercial use: yes
  • Modify and redistribute: yes
  • You must keep: the copyright and license notice
  • Share your changes: no

A summary, not legal advice: the LICENSE file is what applies.

Getting Started

Install Docling with pip:

pip install docling

For installation options and examples, see the README and documentation.

Alternatives

  • xberg: Xberg offers a Rust core with bindings for 15 languages and CLI, REST, and MCP interfaces, while Docling is a Python library.
  • indra: Indra focuses on CPU-based OCR for PDFs and images, while Docling also handles Office files, audio, and other formats.
  • marker: Marker emphasizes Markdown, JSON, HTML, and chunk exports with local models, while Docling offers broader document-format support.
  • unstructured: Unstructured parses documents into structured elements with format-specific dependencies, while Docling also supports audio and local processing.
ProjectLanguageLicenseStarsStatus
doclingPythonMIT69.6kActive
xbergRustMIT9.4kActive
indraPythonApache-2.0421New
markerPythonApache-2.040.2kActive
unstructuredHTMLApache-2.015.5kActive

Considerations

Docling is actively developed, with regular commits and releases. Changes are merged and issues are handled promptly in the reported project activity, and contributions are not concentrated in one person. Its broad format support can make adoption useful across varied input types, but teams should verify the formats and extraction behavior they depend on. Python 3.10 or later is required; Docker support is available, though the README's basic installation path uses pip. The README identifies metadata extraction and complex chemistry understanding as upcoming rather than current capabilities. The code is MIT-licensed, but model licenses need to be checked separately.

Found this useful?

Share it with someone who would like docling.

Comparisons