Google AI has rolled out Gemini Omni 1.2 Flash, making the model generally available across the Gemini API and Gemini for business. The generative AI update introduces native 15-second video creation, multi-turn conversational editing, and enhanced reference following. A 30% price drop from the previous version makes 720p video available at $0.07 per second, positioning the model as fast and cost-effective for large-scale projects.
Key features include longer native clips up to 15 seconds, up from 10 seconds previously, and the ability to specify duration in one-second increments from 3 to 15 seconds. The model offers better reference adherence, allowing users to guide generation using character, product, or style images, short motion clips, and audio cues. Faces, garments, packaging typography, and brand colors maintain consistency across camera angles, facilitating virtual try-on clips and branded video ads.
More consistent results reduce visual drift across frames, whether interpolating between keyframes or extending clips. Easier conversational editing enables turn-by-turn refinements to swap backgrounds, change lighting, or adjust aspect ratio across full 15-second clips. More natural speech capabilities include fluid dialogue, clearer lip-sync, and distinct multi-speaker conditioning for avatars and narrated videos.
Source: Hacker News · Summarized by HeadlinesBriefing