Google has released Gemini Omni 1.2 Flash, a multimodal video generation and editing model from Google DeepMind that produces native 15-second clips at 720p for $0.07 per second—a 30% reduction from Omni 1.1 Flash’s ~$0.10 per second, totaling approximately $1.05 per full clip. The model accepts text, images, short video clips, and audio as input, outputting video with synchronized audio. Key upgrades over 1.1 include extending maximum clip length from 10 to 15 seconds with 1-second duration control (3–15s), improved reference adherence allowing combination of character, product, or style images with up to 3-second motion clips, and enhanced consistency for faces, garments, packaging text, and brand colors across camera angles. Speech synthesis now features more natural dialogue, clearer lip-sync, and distinct multi-speaker conditioning, while camera controls such as FPV flythroughs, zoom-outs, and orbits exhibit fewer unwanted jump cuts. A draft-edit-upscale workflow enables cost control: users draft at 360p or 720p, apply targeted edits via multi-turn conversational editing across the full 15-second span using a previous_interaction_id to avoid re-uploading, then upscale to 1080p or 4K. The model is available via the Gemini API, Google Cloud (with Provisioned Throughput and enterprise SLAs), and Google apps including Flow, Google Vids, Google Ads, and the Gemini app (rolling out to AI Ultra). Outputs carry a SynthID watermark. No benchmark scores for 1.2 are published; Google cites side-by-side demos and partner feedback (e.g., invideo, QuickFrame AI) noting better prompt adherence and physics. Audio references remain allowlist-gated on Google Cloud. Weights are closed; no local deployment is supported.