Gemini Omni, the ‘create anything’ model, starts today with lifelike video
9to5Google • • general
Key Points:
- Google has introduced Gemini Omni, a new generative model family focused on creating highly realistic videos by integrating text, images, video, and audio inputs into unified outputs.
- Gemini Omni builds on Google's previous models like Genie, Veo, and Nano Banana, with an emphasis on maintaining a cohesive, grounded world and offering multimodal understanding.
- The first model, Gemini Omni Flash, is available now to AI Plus subscribers, while a free version will launch soon on YouTube Shorts and YouTube Create; a more advanced "Omni Pro" model is also planned.
- To ensure responsible use, Google will watermark all Gemini Omni videos with SynthID and restrict audio editing capabilities initially, while allowing users to create personalized avatars for their videos.
- Early demos demonstrate impressive realism, including accurate physics and sound effects, showcasing the model’s ability to generate complex, believable video content.