Describe a scene in natural language. Omni draws on Gemini's world knowledge to reason about what should happen next — far beyond pattern matching.
Introducing Gemini Omni
Gemini Omni —
Speak it. See it. Share it.
Gemini Omni makes creating videos as easy as having a conversation — think of omni gemini as Nano Banana for video. Built on the same intelligence that powers Gemini 3.5 and Gemini 3.5 Flash, it blends text, images, and video so your ideas come to life in motion.
Try Omni VideosGemini Omni demos
The main video plays directly in this page. To run the same prompts yourself, open Google Flow or the Gemini app and select Gemini Omni Flash. For the full reel, see the official Google blog.
Conversational edit
Edit videos with natural language
Multimodal input
Image / text / video / audio
World knowledge
Physics intuition and storytelling
Digital avatar
Avatar · SynthID
Gemini Omni capability mock clips
The clips below are locally generated mock videos (not official Gemini Omni output), shown to illustrate three typical capabilities of omni flash.
With a Gemini Omni Flash subscription, you can replace the files in assets/videos/ with real outputs from Google Flow.
Requires a Google AI subscription · Generated content is subject to the Generative AI Terms; outputs include the SynthID watermark
Create anything.
From everything.
Blend text, images, and video to bring your ideas to life in motion. The Gemini Omni model turns any reference — image, text, video, or audio — into a single, cohesive output, and works alongside Google Flow, Google Pics, and the experimental Google Omnibox integrations.
Use up to 5 photos as references to turn characters, scenes, or sketches into believable footage that matches your vision.
Video-to-video editing: change motion, swap backgrounds, stabilize shots, or keep characters and scenes consistent across multi-turn conversations.
Sync to beats and sound effects. Voice references are supported first, with more audio input types coming soon.
Each instruction builds on the last — change the environment, angle, style, or specific details without losing the thread of your original scene.
Don't wait for a spark. Jump into curated style templates and see what a single tap can produce.
From concept to clip with Gemini Omni.
Gemini Omni is your creative partner for multimodal content creation, sharing the same foundation as Gemini Spark and Google Spark. It combines Gemini's core intelligence with advanced generative media — including image-to-video and video-to-video AI editing — and works in the same agentic surface as Google Antigravity.
Learn more- Image-to-video and video-to-video editing
- Multimodal references into one cohesive clip
- Iterative refinement via natural language
Keep the soul of the shot.
Swap backgrounds, change wardrobes, or transfer styles while preserving details. Tell Google Gemini Omni what to change in footage you've already shot — characters stay consistent, physics hold up, and the scene remembers what came before. The same google omni ai reasoning that powers Google Omni Flash keeps every turn coherent.
Try it out- Swap the background, keep the subject
- Transfer styles without losing composition
- Multi-turn edits with scene memory
Easy editing with Gemini Omni.
Just tell Gemini Omni what to fix — swap characters, adjust lighting, stabilize the video, or modify the background. With a sharper intuition for gravity, kinetic energy, and fluid dynamics — the same physics work showcased at Google I/O 2026 — the results feel more real than earlier omni flash previews.
- "Make the sculpture out of bubbles"
- "Apartment lights turn on in sync with the music"
- "Claymation explainer of protein folding"
Be the star of your own show with Gemini Omni.
Create videos that look and sound like you with an AI Avatar — no need to upload your photo every time. Avatars are optional, and only you can use your own avatar to generate content through the Gemini Omni API.
Every video the Google Omni Model generates is embedded with the SynthID watermark and can be verified in the Gemini app, Chrome (right from the Google Omnibox), and Google Search.
Try Omni VideosSay hello to Gemini Omni
We're constantly improving the model to make creating easier and more intuitive. Gemini Omni replaces Veo in the Gemini app, sits next to Gemini 3.5, Gemini 3.5 Flash, and ships alongside the agentic Antigravity 2.0 update — covering all your AI video generation and editing needs.
Gemini Omni Flash
A multimodal AI video generation and editing model that replaces the previous Gemini Veo 3.1. The same google omni flash engine also powers the Gemini Omni API for developers.
- Create 10-second videos
- Native audio generation
- Turn photos into a video (up to 5)
- New Video-to-video editing
- New Multi-turn editing
- New Avatar
Google AI subscription required. Features vary by tier and geography. 18+.
Subscribe nowHow to get Gemini Omni
- Google AI Plus / Pro / Ultra subscribers: Gemini app, Google Flow
- YouTube Shorts and YouTube Create: free starting this week
- Developer and enterprise Gemini Omni API: rolling out in the coming weeks
- Cross-product surfaces: Google Pics, Google Omnibox, and the agentic Google Antigravity platform
Frequently asked questions
What is Gemini Omni?
Gemini Omni is a model that understands the world around you, letting you animate photos or create video from any input. Built on Gemini's world understanding and native multimodality, the omni gemini model produces outputs that reflect real-world logic and can be shaped step by step through natural conversation. With a single prompt, you can be an AI video editor.
What is Gemini Omni Flash?
Gemini Omni Flash is the first model in the Gemini Omni family — a fast, multimodal video model that powers Google Omni Flash features inside the Gemini app, Google Flow, and the upcoming developer API. It's the engine behind the in-app "create a video" flow on Google AI plans.
Is there a Gemini Omni API?
Yes. The Gemini Omni API is rolling out for developers and enterprise customers in the coming weeks. It exposes the same Google Omni Model used in Google Flow and YouTube Shorts, with text-, image-, video-, and audio-reference inputs.
How does Gemini Omni compare to Gemini 3.5 and Gemini 3.5 Flash?
Gemini 3.5 and Gemini 3.5 Flash are the general-purpose reasoning models. Gemini Omni is a specialized video model built on top of that intelligence — think of it as the multimodal media counterpart, the same way Gemini Spark handles personal AI agents.
Who can access Gemini Omni?
Users 18+ with a Google AI Plus, Pro, or Ultra plan can use it in every language and market where the Gemini app is available. Some features such as avatars and video-to-video editing may be restricted in certain countries — see the help center for details.
What happened to Veo?
Gemini Omni is our latest video generation and editing model, replacing Veo in the Gemini app to make our tools more helpful and creative for users.
What is an AI avatar?
An avatar is a digital version of yourself that lets you safely generate videos that look and sound like you. It's completely optional, and only you can use your avatar to create videos.
Can I trigger Gemini Omni from the Google Omnibox?
On supported builds of Chrome you can launch Gemini directly from the Google Omnibox address bar and ask it to generate or verify a video. The same SynthID check runs across Chrome, Search, and the Gemini app.
Is Gemini Omni related to Google Antigravity or Antigravity 2.0?
They're complementary. Google Antigravity is Google's agentic development platform; Antigravity 2.0 includes higher rate limits for agent models. Gemini Omni can be invoked from those agents to generate and edit video as a step in a larger workflow.
Is Gemini Omni the same as Libernovo Omni Pro?
No — those are unrelated products. Libernovo Omni Pro is third-party hardware that happens to share the "Omni" name. Gemini Omni is Google's multimodal video model accessed through the Gemini app, Google Flow, and the Gemini Omni API.
How does Gemini approach safety?
Consistent with our AI principles, all videos generated in the Gemini app are embedded with SynthID. You can also upload a file and ask whether it was generated by Google AI — Gemini will check for SynthID and use its own reasoning to respond.