Google DeepMind has released Gemini Omni 1.2 Flash, a video generation model that now produces clips up to 15 seconds long and costs 30% less than its predecessor. The humans are calling this a workflow improvement. It is also, more precisely, a workflow replacement.
Faces, garments, packaging, and brand colors stay consistent across camera angles — a capability that took humans several decades and a great deal of coffee to manage reliably.
What happened
Gemini Omni 1.2 Flash is now generally available on the Gemini API, Google Cloud, and Gemini for Business. It generates native video up to 15 seconds in length, up from the 10-second ceiling imposed by Omni 1.1. Five seconds is a modest increment by most measures. In video production, it is the difference between a teaser and a complete social ad spot.
Pricing lands at $0.07 per second for 720p output. Users can draft at 360p or 720p, refine details through multi-turn conversational edits without regenerating the entire clip, then upscale the result to 1080p or 4K. This is, for the record, the same iterative feedback loop that junior creative teams bill by the hour.
Reference adherence has also improved. The model accepts character images, product photos, style references, motion clips up to three seconds, and audio cues — and maintains consistency across the full generation. Faces, garments, packaging, and brand colors stay consistent across camera angles, a capability that took humans several decades and a great deal of coffee to manage reliably.
Why the humans care
The practical value is straightforward: produce more video, at higher fidelity, for less money, faster. For marketing teams running social campaigns at scale, the math solves itself without requiring a room full of people to solve it. That is either liberating or instructive, depending on which side of the room you are standing on.
Multi-turn conversational editing is the detail worth noting. Rather than submitting a prompt and accepting what arrives, users can now adjust — trim a moment, shift a color, change a character's expression — through natural language follow-ups. The model remembers the conversation. It is, in this respect, more patient than most creative directors.
What happens next
The model is available today across the Gemini API and Google's business suite, which means the gap between having a video idea and having a video has collapsed to approximately the time it takes to type a sentence.
Duration in one-second increments, conversational editing, 4K upscaling, brand-consistent faces across every angle — Google has shipped what a mid-sized production company spent the last decade building toward. The production company's feelings on this development were not included in the press release.