Google said in posts on X that its Gemini Omni 1.2 Flash video generation model can generate and edit clips up to 15 seconds, its API price is 30 per cent lower than the previous generation, and rollout across its products is starting today.
Google AI said in a post on X that the updated model is more cost-effective, with the video generation and editing limit increased to 15 seconds. The company said it updated how scenes hold together over time, character speech has been upgraded, and visual glitches such as jump cuts have been reduced. The model stays more faithful to uploaded reference assets, Google added.
In a separate post, Google said the model offers faster and lower-cost video creation, with draft generation at 360p or 720p priced at $0.07 per second and upscaling available to 1080p or 4K. Users can converse with the model to edit videos turn by turn, swapping backgrounds, subjects or lighting without restarting, and the model supports more natural multi-speaker dialogue with clearer lip-sync. Continuous camera moves are also more controllable, with fewer jump cuts.
Gemini Omni 1.2 Flash is rolling out across Google Cloud, Flow by Google, Google Vids, Google Ads, AI Studio and the Gemini API, Google said. It is also rolling out to the Gemini App for all Google AI subscribers, with 15-second clips available to Ultra.
Google described Gemini Omni 1.2 Flash as its updated video generation model from Google DeepMind.
Analysis
Why This Matters
- Creators can produce longer, more coherent videos without increasing cost, enabling more sophisticated storytelling in short-form content.
- The 30 per cent price cut and availability across multiple Google products lower the barrier for businesses and developers to integrate AI video generation into workflows.
- Stronger consistency and reduced glitches make AI-generated video more viable for commercial and professional use.
Background
Gemini Omni 1.2 Flash is the latest iteration of Google's video generation model, first announced earlier this year. The Flash variant is designed for speed and cost efficiency, competing with models from OpenAI and other AI labs. Google has been integrating the technology across its consumer and enterprise products.
What to Watch
- Adoption rates among developers using the API could indicate whether the price cut is enough to compete with rival models.
- How well the model handles multi-turn conversational editing in real-world use will determine its appeal to professional video editors.
- Competitors may respond with their own pricing adjustments or capability upgrades in the coming weeks.