Google DeepMind launches Gemini Omni 1.2 Flash with 15-second video clips and 30% lower cost
Gemini Omni 1.2 Flash offers longer 15-second videos, improved consistency, multi-turn editing, and 30% cost reduction for video generation.
Google DeepMind has launched Gemini Omni 1.2 Flash, an updated AI video generation model that enables native creation of 15-second clips, improved consistency in visuals, and multi-turn conversational editing. The new version reduces costs by 30% compared to its predecessor and is now generally available on the Gemini API, Google Cloud, and Google Vids platforms.
Gemini Omni 1.2 Flash allows users to specify video durations from 3 to 15 seconds in one-second increments, up from a 10-second limit in the previous version. This enables the production of more complete stories, product explainers, and social media ads in a single generation. The model supports editing clips through multi-turn conversations, allowing for faster adjustments without regenerating entire videos.
Visual consistency has improved significantly, with better adherence to reference images, stable character and object details across motion and camera transitions, and a more cinematic look with smoother camera movements. The model maintains accurate lip-sync and natural dialogue flow for multi-character scenes.
Gemini Omni 1.2 Flash introduces several enhancements over version 1.1. Users can generate and edit longer clips up to 15 seconds at 720p resolution, with the option to draft at lower resolutions and upscale to 1080p or 4K. The cost is now $0.07 per second for 720p video, representing a 30% price drop.
The model supports detailed guidance combining character, product, and style images, short motion clips, and audio cues. It preserves faces, garments, packaging typography, and brand colours consistently across camera angles. Multi-turn conversational editing enables iterative changes such as swapping backgrounds, adjusting lighting, or changing aspect ratios while keeping core elements intact.
Gemini Omni 1.2 Flash also improves cinematic camera control, enabling continuous drone flights and complex movements without jump cuts. The model holds references more tightly across longer shots, reducing visual drift or morphing. Outputs carry an imperceptible SynthID watermark and are backed by Google Cloud's enterprise infrastructure.
African businesses, content creators, and developers could benefit from Gemini Omni 1.2 Flash’s cost-effective and scalable video generation capabilities.
Google DeepMind plans ongoing development to improve features such as fast sports action, rapid script changes, and complex physics simulation. Users can start using Gemini Omni 1.2 Flash immediately across Google’s developer and business platforms, including Google AI Studio, Google Cloud console, Google Vids, and the Gemini app for subscribers.
Sources
Source: Google DeepMind
