Shengshu Technology has rolled out Vidu 2.0, the latest version of its video generation platform, designed to make content creation faster, easier, and more affordable. The update enables users to generate video clips in as little as ten seconds while cutting costs by more than half compared to industry averages.
Central to Vidu 2.0 is Shengshu’s universal vision transformer (U-ViT) model, combined with its proprietary full-stack interference accelerator. These innovations allow the platform to deliver high-quality videos at speeds and costs previously unattainable in the market.
With industry-standard costs averaging USD 0.084 per second of video, Vidu 2.0 offers a generation rate of approximately USD 0.0375 per second—55% lower than the norm and half the cost of its predecessor, Vidu 1.5.
The update also introduces a new feature offering templates (preset prompts) to simplify video creation. Users can add detailed actions or props with just a few clicks, bypassing the need to manually write prompts. For example, an outfit or setting can be applied directly from the preconfigured library, making the platform accessible even for beginners. Shengshu plans to expand the template library over time, offering users more options for their projects.
Vidu 2.0 represents a step toward Shengshu’s broader vision of enabling AI-driven storytelling that dynamically adapts to creators’ choices.
Founded in March 2023, Shengshu specializes in deep generative models, combining diffusion techniques and transformers for multimodal applications. The company transitioned from research to commercialization with the initial release of Vidu in July 2024. Shengshu also developed VoxCraft, a 3D asset generation platform.
Vidu’s advancements come as generative AI tools gain global traction. Last month, OpenAI’s Sora expanded public access, and Google integrated its latest video model, Veo 2, into its VideoFX tool, which is currently available to select users with a public waitlist. In China, Shengshu faces competition from domestic players such as Kuaishou’s Kling AI, MiniMax’s Hailuo AI, and ByteDance’s Dreamina.