Pyyan / OCR & Document AI / olmOCR

olmOCR

Fully open pipeline, weights and data.

currentAllen Institute for AIVerified 2026-08-13

Open

end to end

Specifications

OmniDocBench~91
Open weightsYes
HandlesTables, markdown structure
LicenceApache 2.0
KindVision language model

What matters

  • Preserves reading order and markdown structure
  • Training data published, which is rare

Best for reproducible research pipelines.