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-gated

ungated(nothing is withheld from the published tree

Two licenses, and the compound is resolved on the more restrictive of them because both cover the artifact you actually run. The code is Apache-2.0. The model weights the pipeline requires - Marker does not convert a document without them - are under a modified AI Pubs Open RAIL-M that the README states is 'free for research, personal use, and startups under $5M funding/revenue', with a commercial license required above that threshold. A license charging for a class of use above a revenue threshold is the competition_restricted tier, and a public source under that tier scores 2/source_available. This is deliberately not the model-context-protocol case, where a CC-BY-4.0 documentation license was moved out of the compound because it covered the project's prose rather than the artifact you run: here the bounded license covers the weights, and the weights are the product.

• https://raw.githubusercontent.com/datalab-to/marker/master/LICENSE recorded 2026-08-30

  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 2026-08-30

  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 2026-08-30

  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.'

• 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.

289,172 downloads of marker-pdf in the trailing 30 days, which bands at 100K-1M, level 3 on the software adoption scale.

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

  last_month = 289,172 downloads for marker-pdf.

Banded one below the docling anchor on the same reading as MinerU: the same layout, table, equation and OCR path over documents, without the audio, video and application-specific XML breadth the top band is anchored on. The comparison is dated to when the anchor was last confirmed rather than to this sweep, because a relative band is only as fresh as the record it is relative to.

• https://raw.githubusercontent.com/datalab-to/marker/master/README.md recorded 2026-08-30

  README documents conversion of PDFs, images, Office documents and EPUB to Markdown, JSON and HTML, with layout analysis, table and equation recognition, OCR and optional LLM-assisted refinement.