OpenOCR provides a unified training and evaluation benchmark for general OCR tasks including text detection, recognition, formula recognition, and table recognition. It integrates commercial-grade document parsing systems and faithfully reproduces implementations from academic papers, using PyTorch as the underlying deep learning framework.