Google Expands Influence in AI-Powered Video Generation with Gemini Omni Models
In Focus
- The Gemini Omni model is multimodal in both input and output
- Gemini Omni Flash is the first model in the Omni family of AI models
- Omni-generated videos will carry the SynthID watermark
Google has launched the Gemini Omni AI model at the I/O developer conference. The new model features advanced AI-powered video generation capabilities. Omni differs from text-to-video tools because it is multimodal in both input and output. This means it supports text, audio, images, and video inputs, and uses Gemini’s real-world knowledge to create a unique, interactive world.
What is Google Gemini Omni?
Gemini Omni is a new family of multimodal AI models designed to “create anything from any input”. The first model in this family is the Gemini Omni Flash, which Google has integrated into the Gemini app, YouTube Shorts, and Google Flow.
According to the search giant, Gemini Omni flash combines reasoning and creative abilities, allowing users to create visuals, starting with video. Like Nano Banana, Omni allows creators to edit photos using simple text commands.
“With Omni, you can combine images, audio, video and text as input and generate high-quality videos grounded in Gemini’s real-world knowledge. You can also easily edit your videos through conversation,” Google DeepMind CTO and Chief AI Architect, Koray Kavukcuoglu noted in a blog post.
Google introduced Gemini Omni months after launching the Gemini 3.1 Flash-Lite AI model for handling high-volume developer workloads.
What Can Gemini Omni Flash Do
Omni Flash makes it easier for users to edit videos using natural language. The AI model maintains consistency across characters and effectively builds new instructions on the previous ones. Using Google Gemini Omni, users can change entire scenes in videos or specific aspects.
The AI model also allows users to reimagine their videos. For instance, a user can record a video and ask Omni to change it. They can add new objects or characters to their videos, or change it into something entirely new. Most importantly, Omni combines Gemini’s knowledge of science, history, and cultural context with physics to create accurate scenes.
“Omni has an improved intuitive understanding of forces like gravity, kinetic energy and fluid dynamics, allowing you to create more realistic scenes,” Kavukcuoglu added.
Users Can Bring Input References to Omni
With Omni Flash, users can input references or natural language descriptions to define video style, motion, or effects to generate cohesive clips. This means they can use character images, drawings, or scenes to create outputs that match their preferences.
Omni also comes with an avatar feature that allows users to create digital versions of themselves. Such actions allow them to generate videos that sound and resemble their appearance. Google said all Omni-generated videos will carry the SynthID watermark.
What is the Future of Gemini Omni?
Google’s long-term vision is to have Omni become the main tool for generating audio from videos and images from audio. “With world models, AI is moving from predicting text to simulating reality. Gemini Omni is the next step in that direction,” Google CEO Sundar Pichai mentioned as cited by TechCrunch.
