PaddleOCR 3.5 has been released with support for running OCR and document parsing tasks using a Transformers backend, according to a blog post from PaddlePaddle on Hugging Face. The post frames the release as a step toward making PaddleOCR's document-processing capabilities available within the Transformers ecosystem.
Because there is only one source, there are no independent details to compare against. The announcement focuses on the new backend integration rather than providing benchmark results or performance comparisons, so the practical impact beyond the stated functionality is not covered in the source.