ShengShu’s Vidu Pushes AI Video Into Production With New Q3 and Q2 Pro Models

New updates have been reported about ShengShu Technology.

ShengShu Technology is accelerating the industrialization of AI video through a major upgrade of its Vidu platform, positioning the company as a key infrastructure provider for production-grade generative video. At its Global Creativity Week in Singapore, ShengShu introduced Vidu Q3, a long-form AI video model that generates synchronized audio and visuals in a single pass and has already ranked No. 1 in China and No. 2 globally in benchmarks by Artificial Analysis.

Complementing Q3, ShengShu launched Vidu Q2 Reference-to-Video Pro, which transforms Vidu from one-shot video generation into a controllable, reference-driven workflow suitable for enterprise and professional studios. Together with Reference Hub 2.0 and the upgraded Vidu Agent, these tools integrate planning, editing, and automation, enabling faster revisions, higher consistency, and scalable creative asset reuse for commercial users.

ShengShu’s focus is explicitly on converting frontier research into measurable productivity, with CEO Yihang Luo emphasizing that model breakthroughs only matter when they embed directly into daily production workflows. Vidu Q3 currently supports up to 16 seconds of native multilingual audio-video output with 1080p resolution, precise lip sync, cinematic camera control, seamless transitions, and text rendered as part of the frame, making it suitable for animation, short-form drama, and film projects.

Vidu Q2 Pro allows up to two video and four image references in a unified pipeline, enabling teams to lock in people, scenes, actions, and visual styles while adding or modifying elements without regenerating entire sequences. This capability is designed to lower production risk, compress revision cycles, and enhance brand consistency, key concerns for advertisers, media companies, and IP owners exploring AI-driven pipelines.

On the ecosystem side, ShengShu introduced Reference Hub as a subject-based creation community built on its reference architecture, offering more than 200 curated subjects across camera movement, composition, narrative structure, visual style, and character performance. This modular approach is intended to standardize high-quality visual components, reduce reliance on scarce top-tier creative talent, and support repeatable, large-scale campaign and content rollouts.

Vidu Agent 1.0 further streamlines workflows with one-click video generation, storyboard-based editing, and support for custom background music and narration removal, effectively acting as an orchestration layer for non-technical users within enterprise teams. Underpinning these capabilities is ShengShu’s U-ViT architecture and its TurboDiffusion collaboration with Tsinghua University’s TSAIL Lab, which reportedly boosts inference speeds by up to 200 times, improving unit economics for high-volume video generation.

Since its April 2024 launch, Vidu has scaled to users in more than 200 countries and regions, serving over 40 million creators and more than 10,000 developers and enterprise customers, with over 500 million videos generated and over 70 percent tied to commercial use cases. Adoption spans creative platforms, AI workflow providers, production houses, and consumer brands, signaling growing trust in ShengShu’s stack as a core component for AI-enabled content and marketing operations.

For executives, the strategic implication is that ShengShu is evolving from a model provider into a full production platform aimed at capturing workflow, data, and recurring spend across the global content value chain. If it sustains benchmark leadership and execution speed, ShengShu is positioned to benefit from rising demand for scalable, controllable AI video in advertising, entertainment, e-commerce, and digital experiences, while creating potential partnership, integration, and competitive considerations for incumbents in media and creative software.