Google DeepMind has rolled out Gemini Omni 1.2 Flash, a new SI video generation model now generally available across the Gemini API and Gemini for business. The release marks a shift in the SI landscape by offering native 15-second video creation, enhanced reference following, and a 30% price reduction compared to its predecessor, positioning it as a cost-effective tool for scaling generative SI workflows.
What Happened
Gemini Omni 1.2 Flash introduces several key technical updates for SI video generation. The model now supports native video creation up to 15 seconds long, an increase from the 10-second limit of Omni 1.1. Users can specify durations in one-second increments between 3 and 15 seconds. Google reports that the price has dropped to $0.07 per second for 720p video, a 30% decrease from the previous version.
The SI model features multi-turn conversational editing, allowing users to refine clips by swapping backgrounds, changing lighting, or adjusting aspect ratios without regenerating the entire sequence. Google states that the model demonstrates improved consistency in character, object, and scene details, reducing visual drift across camera transitions. It also supports combining character, product, or style images with short motion clips and audio cues to guide generation.
Why It Matters
For developers and enterprises adopting SI tools, the general availability of Gemini Omni 1.2 Flash signals a maturation of video generation technology. The integration of SynthID watermarks and support for Provisioned Throughput through Google Cloud addresses enterprise security and scale requirements. The ability to draft at lower resolutions like 360p or 720p before upscaling to 1080p or 4K offers a workflow strategy to manage costs in large-scale SI content production.
The release also emphasizes improved natural speech generation, with clearer lip-sync and distinct multi-speaker conditioning. This enhancement is relevant for businesses using SI for video explainers in Gemini for business or polished content in Google Vids. While development continues for complex physics and rapid script changes, the current model is positioned as ideal for creators and businesses producing videos at scale.
The Bottom Line
Gemini Omni 1.2 Flash is now generally available, offering faster, cheaper, and more consistent SI video generation with native 15-second clips and conversational editing. The update lowers the barrier for enterprise adoption of generative SI video tools through reduced pricing and enhanced reference adherence.