unstale

pypi / pymupdf4llm

pymupdf4llm

What changed in pymupdf4llm 0.2 to 0.3, from its official release notes, upgrade guides and advisories. Sources checked 2026-10-19.

1 breaking5 new APIs and options1 behavior and defaults2 fixes and other

See which of these apply to your code, ranked, in your agent:

npx @unstale/cli init

@unstale/cli on npm

pymupdf4llm 0.35 changes

New APIs and options 4

  • A new parameter calls the supplied ocr_function on every page, skipping the OCR worthiness check; it requires ocr_function.

    0.3.4release notes

  • OCR can combine RapidOCR and Tesseract-OCR for improved results.

    0.3.4release notes

  • A new parameter accepts a callable that OCRs a page and provides it with a text layer.

    0.3.4release notes

  • Support was added for using RapidOCR via a callable plugin.

    0.3.4release notes

Behavior and default changes 1

  • The default DPI for OCR changed from 400 to 300.

    0.3.4release notes

pymupdf4llm 0.24 changes

Breaking changes 1

  • Removed the OpenCV dependency; numpy is now used to check whether a page is worthwhile to OCR.

    0.2.9release notes

New APIs and options 1

  • Added a top-level function to extract field names and values from Form PDFs; it is available whether or not PyMuPDF-Layout is active.

    0.2.9release notes

Fixes and other changes 2

  • Fixed image saving when the document filename contains folder specifications.

    0.2.9release notes

  • Page chunk output may fail for erroneous layout bounding boxes; this is fixed.

    0.2.9release notesrelease notes