OvisOCR2 is a 0.8B document parsing model designed as an end-to-end parser. Given a document page image, it generates a Markdown representation in natural reading order, covering text, formulas, tables, and visual regions. We built a data engine that combines filtered real-document annotations with synthetic pages, where the rendered images and Markdown targets are derived from the same HTML source.
The training recipe includes supervised fine-tuning, reinforcement learning on a 4B branch with a multi-component reward design, on-policy distillation into the 0.8B model, and model fusion. On OmniDocBench v1.6, OvisOCR2 achieves a state-of-the-art overall score of 96.58, placing an end-to-end model at the top of this leaderboard previously dominated by pipeline methods, highlighting the potential of end-to-end document parsing. On PureDocBench, OvisOCR2 also achieves the highest Avg3 score of 75.06.
Beyond these two public benchmarks, we evaluate OvisOCR2 on an in-house benchmark designed to cover a broader set of long-tail and challenging scenarios. OvisOCR2 obtains the best overall performance among the compared methods, providing further evidence of its generalization and robustness. The model is available at Hugging Face.
Blogger's Review: The design and training approach of OvisOCR2 showcase the immense potential of deep learning in complex document parsing, particularly through its end-to-end strategy that significantly enhances the model's generalization and robustness. Its superiority in multiple benchmark tests marks a new breakthrough in document parsing technology.