Daily 9/10: Suno's licensed v6, Prime Video AI lip sync, Kakao's KATok

Suno has launched a licensed v6 generation trained with music-industry partners, while Prime Video is applying AI lip sync to a scripted series. Kakao has unveiled KATok, a method for reducing token computation in video generation.

• [TechCrunch] Suno Retires Its Existing Models for a Licensed v6 Generation ↗

  Suno has globally launched v6, v6 Wild, and v6 Mini, developed with Warner Music Group, BMG, Believe, and other partners, while retiring its previous generation of models. The new system can use text, images, and video as references for music generation and can isolate a specific instrument from a sample to build a new beat. Suno also plans official remix and derivative-music features in which artists can opt in and receive compensation. The move shifts Suno's core model from a system facing unlicensed-training disputes toward a structure built around music-industry licensing.

• [TheWrap] Prime Video Brings AI Visual Dubbing to a Scripted Series ↗

  Amazon Prime Video is using AI and visual effects to adjust actors' mouth movements to match translated dialogue recorded by human voice performers. The lip-sync technology is being applied to the English dubs of the German series Maxton Hall seasons 1 and 2, and will also be used for season 3 in December. Amazon plans to expand it to additional titles. AI dubbing is moving beyond translated audio and voice generation into frame-level changes that automate part of international content localization.

• [DigitalToday] Kakao Unveils KATok to Cut Computation in AI Video Generation ↗

  Kakao has presented KATok at ECCV 2026, a video-generation method that dynamically adjusts the number of tokens according to the complexity of each video. In experiments, it increased video-model training speed by about 6.9 times and generation throughput by about 3.2 times while limiting quality loss. Kakao plans to extend the technology to high-resolution and longer-form video generation. By reducing token computation in proportion to content complexity, KATok directly targets the speed and GPU-cost structure of commercial video generation.