Physix Frontier badgePhysix Frontier
PublishChinese | EN

Physix Frontier · News Briefing Card (Leiphone · Jul 22, 2026)

Shengshu Tech launches general world model for video and embodied AI

KEY FACTS

  • Zhu Jun proposes a general world model that unifies digital content generation and physical action execution.
  • The company launched three products—Vidu, Vidu S1, and Motubrain—covering the full pipeline.
  • It adopts a MoT architecture that integrates environment understanding, state prediction, and action trajectory generation.
  • It builds a data pyramid system spanning internet video to real robot-collected trajectories.

KEY DATA

42FPSVidu S1 max frame rate
540PVidu S1 output resolution

PHYSIX OBSERVATION

Shengshu Tech is trying to break down the barriers between video AI and the robotics industry, defining the world model as the foundation connecting the virtual and the real. Its core logic is to use massive video data to pretrain general cognition, then calibrate physical precision through robot data. If this "digital trial-and-error, physical verification" loop can run successfully, it will not only reshape the content production paradigm but may also provide a low-cost, highly generalizable training path for embodied intelligence. The industry should closely watch the stability of its deployment.

Source: Leiphone report