SINGAPORE, Jan. 30, 2026 /PRNewswire/ -- ShengShu Technology, a global leader in multimodal generative AI, unveiled a series of major upgrades to its AI video platform, Vidu, during its week-long Global Creativity Week, underscoring its push to move AI video from experimentation into large-scale production use.

The releases are led by Vidu Q3, the industry's first long-form AI video model to deliver native audio and video generation in a single output. Designed for narrative production, Q3 integrates sound and visuals directly at the model level, marking a shift from silent visual generation to fully synchronized storytelling. In the latest rankings released by the AI benchmarking authority Artificial Analysis, Vidu Q3 ranked No.1 in China and No.2 globally. The model placed ahead of several leading international video generation platforms, positioning Vidu among the world's top-tier solutions.

Alongside Q3, the company introduced Vidu Q2 Reference-to-Video Pro, the world's first video model where anything can serve as a reference. By expanding AI video from one-shot generation into structured reference-driven workflows, Q2 Pro enables greater control, consistency, and production efficiency. Reference Hub 2.0 and an upgraded Vidu Agent further extend the platform into an integrated production system designed for professional creators and enterprise teams.

At a time when global debate continues over whether AI breakthroughs are translating into sustainable productivity, Vidu's strategy reflects what it calls "China Speed" — the rapid conversion of frontier research into deployable systems embedded directly into production workflows.

"Advanced AI only matters when it creates measurable productivity," said Yihang Luo, CEO of ShengShu Technology. "Our focus is not just model capability, but how quickly that capability becomes usable in daily production."

From AI Capability to AI Productivity: First-of-Its-Kind Production Breakthroughs

During Global Creativity Week, Vidu unveiled a tightly integrated suite of upgrades aimed at transforming AI video from experimental capability into production-ready infrastructure.

Vidu Q2 Reference-to-Video Pro marks a decisive step toward controllable creation. As the world's first video model where anything can serve as a reference, it supports two video references and four image references within a unified workflow. Creators can combine inputs across people, scenes, actions, expressions, effects, and textures, then add, remove, or modify elements without regenerating an entire sequence. By shifting from one-shot generation to reference-driven editing, Q2 Pro turns AI video into an iterative, collaborative tool suitable for professional production environments where revision speed and consistency matter.

Building for storytelling, Vidu Q3 advances synchronized audio-video generation directly at the model level. It is the industry's first long-form AI video generation model to support up to 16 seconds of native audio-video output in a single pass. Sound and vision are generated together directly from the model, delivering synchronized output in high definition. Q3 supports multilingual voice generation, precise lip synchronization, cinematic camera control, seamless shot transitions, and native 1080p rendering, with text generated as part of the visual composition rather than added in post-production. These capabilities position Q3 for animation, short drama, film production, and other narrative-driven formats.

Vidu also introduced Reference Hub, the world's first AI video subject-based creation community. Built on Vidu's reference-driven video generation framework, it enables creators to select from eight structured subject categories to produce professional-grade videos with greater consistency and efficiency. At launch, more than 200 curated subjects are available across camera movement, composition, narrative structure, visual style, scene design, character performance, abilities, and mood. By standardizing and modularizing visual elements, the platform lowers the barrier to high-quality output while enabling scalable creative asset reuse — introducing a new paradigm for AI-powered video creation.

To further streamline production workflows, Vidu Agent 1.0 enables one-click video creation powered by intelligent planning and real-time generation. The upgraded agent supports custom background music uploads, narration removal, and storyboard-based editing, reducing coordination complexity and accelerating end-to-end production.

Research Depth, Engineering Speed

Vidu's rapid product cadence reflects ShengShu Technology's engineering-driven approach. Its earlier U-ViT architecture pioneered the integration of diffusion and transformer frameworks, laying an early foundation for modern video generation models. In late 2025, the company co-released TurboDiffusion with Tsinghua University's TSAIL Lab, significantly accelerating inference up to 200 times while maintaining generation quality.

Over the past year, Vidu has rolled out successive model upgrades, achieving benchmark-leading performance in text-to-video generation and expanding high-consistency reference-based creation. These advances culminate in the reference-driven precision of Q2 Pro and the native audio-video integration of Q3.

Expanding Global Adoption

Since launching in April 2024, Vidu has expanded to users in more than 200 countries and regions, serving over 40 million creators and more than 10,000 developers and enterprise customers. More than 500 million videos have been generated on the platform, with commercial projects accounting for over 70 percent of total output.

Across industries, creative platforms such as PhotoGrid have embedded Vidu's capabilities into their core offerings. AI workflow platform Pollo AI applied Vidu's image-to-audio-video system to address long-standing synchronization challenges. Production teams have used Vidu in long-form narrative projects, while fashion brand Odin deployed the technology to power AI-driven virtual try-on experiences.

"AI becomes meaningful when it is used at scale in real workflows," Luo said. "Global Creativity Week is another step in transforming AI video innovation into practical creative infrastructure."