Marker

license

Apache-2.0(OSI, over the code)+AI-Pubs-Open-RAIL-M-Modified(model weights

source

public(the repository builds the conversion pipeline the package installs)

core features withheld

no — nothing is withheld from the published tree

Marker's code is Apache-2.0, but the model weights it requires to run are under a modified AI Pubs Open RAIL-M license, free for research, personal use and startups under $5M in funding or revenue, with a commercial license required above that. Because the more restrictive license covers the weights actually run rather than just the code, this reads as source-available rather than fully open.

• https://api.github.com/repos/datalab-to/marker recorded 2026-08-30

  Repo metadata - license spdx_id Apache-2.0, archived false, 39,389 stars - for datalab-to/marker.

• https://raw.githubusercontent.com/datalab-to/marker/master/LICENSE recorded 2026-10-07

  LICENSE body is the Apache License 2.0 text, covering the repository's code.

• https://raw.githubusercontent.com/datalab-to/marker/master/MODEL_LICENSE recorded 2026-10-07

  MODEL_LICENSE body is the AI PUBS OPEN RAIL-M LICENSE (MODIFIED), Version 0.1, applying use restrictions to the model weights and their derivatives.

• https://raw.githubusercontent.com/datalab-to/marker/master/README.md recorded 2026-10-07

  README, Commercial usage section: 'Our code is licensed under Apache 2.0 - free to use, including commercially. Our model weights use a modified AI Pubs Open Rail-M license (free for research, personal use, and startups under $5M funding/revenue). For commercial use of the model weights beyond that, visit our pricing page.'

Adoption is measured as PyPI downloads of the marker-pdf package over the trailing month.

• https://pypistats.org/api/packages/marker-pdf/recent recorded 2026-08-30

  last_month = 289,172 downloads for marker-pdf.

Marker's layout, tables, equations and reading order all come through into Markdown and JSON. It stops short of Docling, whose extractor also returns the fields a user defines in a template; Marker's converters rebuild the whole document or only its tables, with no step for returning user-defined fields.

• https://raw.githubusercontent.com/datalab-to/marker/master/README.md recorded 2026-09-29

  The raw README states "Marker converts documents to markdown, JSON, chunks, and HTML quickly and accurately", "Converts PDF, image, PPTX, DOCX, XLSX, HTML, EPUB files in all languages" and "Formats tables, forms, equations, inline math, links, references, and code blocks", and lists its converters as "One of `marker.converters.pdf.PdfConverter` (default) or `marker.converters.table.TableConverter`".