Vidu Q4 Preview puts AI acting and 4K video first

What happened

ShengShu Technology, the Beijing company behind the Vidu video generator, released Vidu Q4 Preview on 7 October 2026. It is an early, public version of Q4, the model that will replace Vidu Q3 as its flagship.

The company is shipping the preview before the full release so creators can use it on real projects and shape the final model. Its main promise is better acting: AI characters whose faces, emotions, movements and voices stay in step with each other, aimed at short dramas, advertising and film.

The key details

• Consistent characters: Reference-to-Video takes 1 to 15 image references and up to 3 audio references, so the same person, outfit and voice can carry across shots.
• Performance: expression, emotion, body movement and speech are generated together rather than layered on afterwards.
• Camera and effects: the camera is designed to follow fast action such as fights and chases more smoothly, and effects like explosions and particles blend better into the scene.
• Length and quality: clips of up to 16 seconds, at 540p, 720p, 1080p, 2K or 4K, with 10-bit colour at the higher resolutions.
• Price: launch promotional pricing starts at $0.014 per second. ShengShu's own release lists standard rates from $0.045 per second at 540p up to $0.39 per second at 4K, with 30% off image-to-video and reference-to-video until 30 November. Final pricing may change.
• Where: on the Vidu web app and through the Vidu API platform for developers.

The background

ShengShu was founded in Beijing in 2023 by a team with roots at Tsinghua University, and its Vidu models are one of China's main rivals to Kling, Seedance and Google's Veo. Vidu Q3, released in January 2026, added synchronised sound and video in clips up to 16 seconds and became known for its reference-to-video feature.

Q4 arrives during a busy few months for AI video. ByteDance's Seedance 2.5, Alibaba's Wan 3 and MiniMax H3 all launched over the summer, and Kuaishou has said the full Kling 4.0 will launch this month. Most of these releases compete on length and resolution; Vidu is putting acting and character consistency at the centre instead.

What it means for brands

The hardest part of AI video ads is not a pretty shot, it is a believable person who looks, moves and sounds the same in every scene. A model built around references for face, wardrobe and voice makes recurring brand characters, spokespeople and AI UGC-style creators far more practical.

4K output with 10-bit colour also matters for anything going beyond social, such as TV, outdoor screens or work that needs proper colour grading. It is still a preview, so we would test it alongside the tools already in a production pipeline rather than switch overnight. If you want to see how consistent AI characters work in practice, have a look at our AI UGC videos.