On Oct. 29, Baidu open-sourced Wenxin Unlimited OCR 2, a document-parsing model with 3 billion total parameters and only 500 million active parameters, released under the permissive MIT license on GitHub and Hugging Face.
The model ranked first on the MarathonDocBench and MPDocBench long-document parsing benchmarks and supports unlimited-page document parsing, a notable advance for enterprise document AI.
The release continues a wave of Chinese open-source model drops, with Baidu emphasizing practical, low-cost deployment for real-world document workflows.
Industry observers note that MIT licensing lowers adoption barriers for commercial users, accelerating diffusion of Chinese document-AI technology.
Image: AI-related computer chips. Source: Wikimedia Commons (CC BY-SA 4.0).
