Shengshu Technology launches Vidu S1, real-time video calls from a single image

生数科技发布 Vidu S1，推动视频生成迈向"实时交互"新时代

Shengshu Technology unveiled Vidu S1 at the 2026 Global Digital Economy Conference. The core pitch is real-time interaction: you can video-call an AI character and steer the visuals with voice commands, with no time limit. It uses an autoregressive diffusion approach, predicting subsequent frames from the generated video and voice input—no traditional modeling needed. One image is enough to create a character with a custom voice. It runs at 25 FPS at 540p (up to 42 FPS), with TurboDiffusion cutting compute costs. Closed beta is live, but the post doesn't disclose latency or real conversation quality.

Why it matters: Shengshu Technology dropped a real-time interactive video generation product, shifting from short-clip generation to video-call-style conversation, with technical details and performance numbers provided. Not scoring higher because only launch info is available so far — no thi...