PaddleOCR-VL: Boosting Multilingual Document Parsing via a 0.9B Compact VLM (huggingface.co)

Original title: PaddleOCR-VL: Boosting Multilingual Document Parsing via a 0.9B Ultra-Compact Vision-Language Model

Direct link to PDF: https://ernie.baidu.com/blog/publication/PaddleOCR-VL_Techni...

Baidu claims state of the art performance on their own OmniDocBench (although some recent models like GPT-5 and Qwen3 are not evaluated) and strong results on olmOCR-Bench and Ocean-OCR-Bench.

Wow is this commercially usable? I think the Microsoft and IBM top class document parsing is non commercial use only.