Gemini Omni 1.2 Flash

Video generation costs decrease by 30%, and clip length extends to 15 seconds, enabling more complex narratives.

Google DeepMind has released Gemini Omni 1.2 Flash, an updated AI model for video generation. This new version supports native 15-second video clips, an increase from the previous 10-second limit. It also introduces multi-turn conversational editing, allowing users to refine generated videos through iterative prompts without starting over. The model offers improved reference adherence, ensuring greater consistency of characters, products, and styles across frames, even with complex camera movements. Additionally, Gemini Omni 1.2 Flash provides more natural speech and lip-sync capabilities for generated dialogue and features a 30% cost reduction compared to its predecessor, making it more accessible for professional projects and large-scale content production. These enhancements are now available through the Gemini API, Google Vids, and Gemini for business.

The release of Gemini Omni 1.2 Flash signifies a move towards more practical and cost-effective AI-powered video creation for businesses and developers. By enabling longer video clips and offering more granular control through conversational editing, Google is addressing key pain points in AI video production, such as repetitive generation and the need for extensive post-production. The focus on improved consistency and reference adherence suggests a growing maturity in generative AI's ability to maintain specific brand identities and product details, crucial for advertising and marketing applications.

This update is poised to impact content creators, marketing teams, and developers by lowering the barrier to entry for professional-quality video production. The 30% cost reduction, coupled with enhanced editing workflows, means that organizations can experiment more freely and scale their video content output more efficiently. The improved consistency in visual elements will also be a significant draw for brands that require tight control over their visual identity, potentially reducing reliance on traditional video production methods for certain types of content.