Google DeepMind Launches Gemini Omni 1.2 Flash with 15-Second Video Generation
Frontier AI models · AI creation
Google DeepMind announced the general availability of Gemini Omni 1.2 Flash, supporting native video creation up to 15 seconds with conversational multi-turn editing and a 30% price reduction to $0.07 per second for 720p.
Google DeepMind has introduced Gemini Omni 1.2 Flash across the Gemini API and Gemini for business. The updated model expands native video generation up to 15 seconds, compared to 10 seconds in Omni 1.1, and allows duration selection in one-second increments between 3 and 15 seconds. Pricing is reduced by 30% from the previous version to $0.07 per second for 720p output. The model incorporates multi-turn conversational editing to modify backgrounds or lighting without full regeneration, and supports multimodal reference conditioning covering images, audio, and short motion clips. Outputs include SynthID watermarking.
Key takeaways
- Native video duration increases to up to 15 seconds with duration controls from 3 to 15 seconds.
- Pricing drops 30% relative to Omni 1.1, set at $0.07 per second for 720p resolution.
- Multi-turn conversational editing allows refining elements such as backgrounds or lighting across 15-second clips without full re-generation.
- Reference adherence enables conditioning on style images, short motion clips, and audio cues, supporting continuous camera movements like FPV drone flights.
Source：Google AI Read the original ↗
Prepared, checked and automatically published by AI; not individually reviewed by a human.