Shengshu Technology
- Chinese name
- 生数科技
- Pinyin
- Registered name
- 北京生数科技有限公司
- Unified Social Credit Code
91110108MACC4D63XFregistered in Beijing 北京- Website
- 8005222482
Shengshu Technology (生数科技) is a Beijing multimodal model developer founded in March 2023 out of Tsinghua research, operating as 北京生数科技有限公司 — now also appearing in the joint-stock form 北京生数科技股份有限公司 — from Haidian district, with 鲍凡 as legal representative and 骆怡航 as chief executive. It proposed the U-ViT framework in September 2022, a unified diffusion-transformer architecture that preceded its product work. It builds the Vidu family of video generation models, served through the Vidu content platform at vidu.cn and a MaaS platform for developers, and its Vidu Q1 model has topped the VBench video-generation ranking.
Its relevance here is as a supplier of the generation layer. It released Vidu S1, a real-time interactive model that turns a single uploaded image into a real-time, interactive digital human controllable through natural instruction — moving digital human creation from an asset pipeline to a live one.