Shengshu releases Vidu Q4 Preview: flagship video model from RMB 0.09 per second, debuting #3 on Artificial Analysis image-to-video

Why it matters

【Key points】

Beijing-based Shengshu Technology released Vidu Q4 Preview, the first preview of its next flagship video model (live on its official X account late on Oct 7 Beijing time; formally announced in China on Oct 8).

The release focuses on character performance, camera language and complex effects. Reference-to-video takes up to 15 reference images and 3 reference audio clips to keep characters and voices consistent; it handles explosions, fireworks and particle effects, and outputs from 540P up to 2K and 4K with 10-bit support. Launch pricing starts at RMB 0.09 per second (from $0.014/sec overseas); Shengshu says the same budget yields up to about 5x more content at comparable specs.

Artificial Analysis says Vidu Q4 Preview debuted at #3 on its AA-Video image-to-video leaderboard, up from #19 for Vidu Q3 Pro. The model is also live on inference platforms such as WaveSpeed and fal. Shengshu says it is still a preview and will iterate toward a full release based on creator feedback, with the Q series serving as the base for its real-time interactive and embodied-AI lines.

【Verifiable】

Vidu official X: https://x.com/ViduAI_official/status/2107856382263505219

Beijing Business Today (via Sina): https://finance.sina.cn/2026-10-08/detail-iniunvxf4405008.d.html

Artificial Analysis ranking post: https://x.com/ArtificialAnlys/status/2107987818526146964

【Our read】

1. Price is the headline. Flagship video generation at cents per second makes generating many takes and picking the best affordable for small teams, which matters more to Chinese video-model competition than any single benchmark.

2. The #3 spot is one image-to-video leaderboard for a preview; the full release and text-to-video results are still to come.

3. The video race is shifting from who renders best to who can deliver usable footage cheaply and consistently.