Alibaba open-sources 0.8B document parsing model OvisOCR2

Alibaba Cloud's open-source document parsing model, OvisOCR2, achieved a score of 96.58 on the OmniDocBench v1.6 leaderboard, becoming the first end-to-end model to surpass traditional pipelined methods and reach the top. The model has only 0.8B parameters, trained on Qwen 3.5 after a reduction of 0.8B, and utilizes a complementary engine of real and synthetic data and a multi-stage training process to achieve single-step Markdown output of text, formulas, tables, and other content.