GoogleAnalysisGoogle DeepMind launches Gemini Omni 1.2 Flash with 15-second generation
What happened.Google DeepMind released Gemini Omni 1.2 Flash into General Availability, introducing native video generation up to 15 seconds alongside conversational clip editing.
Why it matters.Early video models produced isolated short clips that required total regeneration for minor adjustments, inflating inference budgets. Combining longer continuous shots with conversational delta editing unlocks practical video interfaces for interactive software. Engineering teams should benchmark editing latency to evaluate automated workflows.
The technical specifications:
- Extended duration: The model generates continuous video clips from 3 to 15 seconds, increasing maximum generation length by 50 percent.
- Reduced pricing: Output pricing dropped 30 percent to 7 cents per second for 720p resolution on the Gemini API.
- Conversational editing: Developers can alter camera angles, replace assets, and adjust lighting through text prompts without regenerating entire videos.
- Visual consistency: Conditioning on up to 6 reference assets preserves character geometry and motion fidelity across multi-shot sequences.
- Developer deployment: The release is available across Google AI Studio, Vertex AI, and Google Vids with SynthID watermarking.