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

PaddlePaddle released PaddleOCR 3.5, enabling supported PaddleOCR models to run with Hugging Face Transformers as an inference backend via the engine="transformers" parameter. The release includes OCR model PP-OCRv5 and document parsing model PaddleOCR-VL 1.5. Developers can configure backend-specific options through engine_config. A live demo is available on Hugging Face Spaces. The release aims to reduce integration friction for RAG and Document AI workflows built on Transformers infrastructure.