PaddleOCR 3.5: Running OCR and Document Parsing Tasks with a Transformers Backend

PaddleOCR 3.5 now supports Hugging Face Transformers as an inference backend for its OCR and document parsing models. This allows developers to run models like PP-OCRv5 and PaddleOCR-VL 1.5 within a Transformers-centered stack by simply setting the engine parameter. The integration aims to reduce friction in document ingestion for applications like RAG, Document AI, and agent-based systems. By making it easier to turn complex documents into structured data, it improves the reliability of downstream LLM workflows. The release also includes options for backend configuration and provides a quick-start guide for implementation.