SignalTech/AI
China's top 3 video models begin to diverge and compete on separate paths
China's top 3 video models diverging signals intensifying competitive dynamics in AI video.
Context
What happened: China's top video generation models are diverging. Kling announced Kling 4.0 on September 28, and Shengshu released Vidu Q4 Preview on October 8, with Seedance, Kling and Vidu forming a leading group.
Key numbers: Vidu Q4 Preview from 0.09 yuan per second. MaaS image to video 720P at 0.59 yuan per second, 1080P at 0.75 yuan per second, 4K at 2.43 yuan per second.
Why it matters: Falling per second costs lower the barrier for creators and small teams, shifting competition from technical benchmarks toward production efficiency and affordable iteration.
What changed
China's top video generation models are diverging. Kling announced Kling 4.0 on September 28, and Shengshu released Vidu Q4 Preview on October 8, with Seedance, Kling and Vidu forming a leading group.
Why it matters
Falling per second costs lower the barrier for creators and small teams, shifting competition from technical benchmarks toward production efficiency and affordable iteration.