Google has launched Gemini Omni, its latest AI model built to push video creation into a new era.
After Nano Banana helped users create and edit images through Gemini last year, Google is now expanding that creative power into video.
The new model is designed from the beginning as a multimodal AI system, meaning it can work with different types of input instead of relying on text alone.
Create video from almost anything
Gemini Omni combines Gemini’s reasoning ability with stronger creative generation.
Users can mix images, audio, video, and text as source material.
The model can then generate high-quality videos based on Gemini’s understanding of the real world.
This makes it possible to turn many different ideas, references, and materials into a single video output.
Natural language video editing
One of the biggest highlights is easier video editing through normal conversation.
Users can give instructions in natural language, then continue refining the video through follow-up prompts.
Each new instruction builds smoothly from the previous one.
This means users can adjust scenes, change environments, or edit specific parts without needing traditional editing tools.
More consistent characters and realistic scenes
Gemini Omni is also designed to keep characters consistent across video scenes.
The model can remember what happened earlier and apply that context to the next changes.
It also understands basic physical rules such as gravity, kinetic energy, and fluid dynamics.
That helps generated scenes look more natural and believable instead of random or disconnected.
Multiple references in one video
The model can combine many types of references at once.
Users can provide images, text, video, or audio as creative input.
During the initial rollout, audio support begins with spoken voice.
Google plans to support other audio types in the future.
This could make Gemini Omni useful for creators who want to build videos from rough concepts, existing footage, sketches, voice clips, or written ideas.
Safety and responsible rollout
Google says safety remains a major priority.
Early users can create videos using their own voice through Avatars, which can build a digital version that looks and sounds like the user.
However, voice and speech replacement tools are still being tested before wider release.
Google says it wants to make sure these features reach users safely and responsibly.
SynthID watermarking included
Every video created with Gemini Omni will include an invisible SynthID digital watermark.
This watermark cannot be seen by the naked eye, but it can support transparency and detection.
Google says verification will work through the Gemini app, Chrome, and Google Search.
Availability
The first model in the family is Gemini Omni Flash.
It is now available worldwide for Google AI Plus users at $7.99 per month, as well as Pro and Ultra subscribers.
It will also be available for free through YouTube Shorts and the YouTube Create app starting this week.
Developers and enterprise customers will gain API access in the coming weeks.
THIS IS our take
Gemini Omni sounds like Google’s biggest creative AI jump yet. If it can turn mixed inputs into editable videos while keeping characters, physics, and story flow consistent, creators may soon treat video editing less like software work and more like directing through conversation.
 Origin: Google
