Youtu-Parsing-Omni统一四类模态解析
Youtu-Parsing-Omni 是一款 50 亿参数解析模型，可将文档、图像、音频和视频转换为具有时空依据的结构化记录。它采用共享权重编码器、跨模态融合、统一输出模式及覆盖十项解析任务的自我演化后训练流程。作者报告其在 OmniDocBench v1.6 的 Overall 分数为 96.96，在 OmniParsingBench 的平均分为 75.08，并已公开代码和模型权重。
该项目以统一的 50 亿参数架构和公开权重覆盖多类模态，为需要跨模态读取、定位和引用证据的智能体提供了解析基础。
Computer Science > Computer Vision and Pattern Recognition
arXiv:2610.39301 (cs)
[Submitted on 29 Oct 2026]
Title:Youtu-Parsing-Omni: One Encoder, One Schema, All Modalities
Authors:Yunhang Shen, Zhangquan Chen, Xiaotian Li, Zhongpeng Cai, Qing Zhou, Zekai Zhang, Kun Zhang, Zezhong Wu, Jian Hu, Haihan Gao, Heting Gao, Xin Li, Yinsong Liu, Deqian Jiang, Di Yin, Haoyu Cao, Xing Sun, Bing Liu, Yanqiu Qu, Yunfei Wu, Kun Yin, Zhixiang Wei, Yongxiang Hua, Qianyu Li, Xiaoxing Hu, Peiyuan Zhang, Lijiang Li, Dong Xu, Zhenchuan Yang, Yiliang Jiang
View a PDF of the paper titled Youtu-Parsing-Omni: One Encoder, One Schema, All Modalities, by Yunhang Shen and 29 other authors
#Youtu-Parsing-Omni#Youtu-Omni-Encoder#OmniSchema#OmniSchema-Aware On-Policy Distillation#Tencent#OmniDocBench v1.6#OmniParsingBench