New Features of Gemini Omni 1.2 Flash
Google has launched Gemini Omni 1.2 Flash, introducing support for native 15-second video clips, multi-turn conversational editing, and improved adherence to visual and audio references. This update reduces costs by 30% and is now generally available through the Gemini API and Google Cloud.
Extended Clip Length and Editing Capabilities
The new version extends the native clip length from 10 to 15 seconds, allowing users to specify durations in one-second increments from 3 to 15 seconds. This flexibility enables the creation of complete stories, product explainers, and social media ads in a single generation.
Cost Management and Workflow Efficiency
Developers can draft videos at lower resolutions of 360p or 720p, apply targeted follow-up edits, and upscale selected outputs to 1080p or 4K. This approach helps manage workflow costs effectively.
Improved Reference Adherence
Gemini Omni 1.2 Flash enhances reference adherence by allowing users to combine character or product images, short motion clips (up to three seconds), and audio cues. This ensures consistency in faces, garments, packaging typography, and brand colors across different camera angles.
Enhanced Continuity and Motion
The model aims to reduce visual drift and unwanted jump cuts, providing steadier scale during complex camera movements. It also delivers clearer lip-sync and distinct multi-speaker conditioning for more natural-sounding dialogue.
Pricing and Availability
The price for Gemini Omni 1.2 Flash has been set at $0.07 per second for 720p video, reflecting a 30% reduction compared to Omni 1.1. The model is generally available on the Gemini API and Google Cloud, and is rolling out across Google Vids, Gemini for business, Flow, and the Gemini app for Google AI Ultra subscribers.
Future Development Priorities
Google is focusing on ongoing development to enhance performance in areas such as fast sports action, rapid script changes, and complex physics. Developers can access resources through Google AI Studio, starter apps, and developer documentation, while businesses can utilize the model via the Google Cloud console.
Original source: Google AI