15 Seconds, 7 Cents per Second: Gemini Omni 1.2 Flash Makes the Video-Generation Math Clearer

Google AI announced Gemini Omni 1.2 Flash on X. The news was reposted by Sander Dieleman. There are four core changes: the limit for video generation and editing is raised to 15 seconds; temporal continuity across scenes has been updated; characters speak more naturally; and visual glitches such as jump cuts are reduced, while results stay closer to the uploaded reference material. The price also dropped: in the API, 720p video costs $0.07 per second, 30% lower than the previous generation.

For ordinary users, there are three entry points. @FlowbyGoogle is for directing scenes and multi-turn editing; Google Vids can generate 15-second clips at no extra charge; and @GeminiApp can directly make 15-second videos. The official demo gives a specific example: using a single style frame to generate a 15-second, four-shot macro science explainer of a living beehive.

Fifteen seconds is not a long video, but multi-shot capability, object anchoring, and dialogue naturalness determine whether you can avoid rework. With per-second pricing, a 15-second 720p clip costs about $1.05, making it easier for developers and enterprises to try it in their workflows. Video models have often been used for demos; what really blocks deployment is shot consistency and per-second cost. This time, both have moved.

Official demo:

Original post: https://x.com/sedielem/status/2118402068423844265