Following its massive showcase at Google I/O 2026, Google is moving at breakneck speed. The tech giant has officially announced that Gemini Omni, its next-generation natively multimodal AI engine built by Google DeepMind, is leaving its subscriber-only sandbox.
The first model in this powerhouse family—Gemini Omni Flash—is officially rolling out to everyone inside Google Flow. Even better? You don’t need a premium Google AI subscription to start creating. Google is introducing a daily free tier inside Flow so anyone can experience what Chief AI Architect Koray Kavukcuoglu calls the intersection of “Gemini’s ability to reason meets the ability to create.”
Watch Google DeepMind’s official demonstration of Gemini Omni’s natively multimodal framework.
The Evolution of “Nano Banana” for Video
Last year, Google introduced Nano Banana, bringing advanced intelligence to conversational image generation and editing. Gemini Omni Flash takes that exact ground-up multimodality and applies it to fluid, moving video.
Rather than just predicting pixels like standard text-to-video generators, Omni reasons through real-world history, science, and cultural context by uniting three core DeepMind architectures: the Gemini reasoning engine, the Veo video rendering backbone, and the Genie world simulation layer. Instead of wrestling with complex editing timelines or traditional keyframes, you can edit raw footage or build entirely new cinematic clips simply by talking to the model.
Key Features & Visual Demos From the Labs
1. Multi-Turn Conversational Video Editing
With Gemini Omni, every single text instruction builds organically on top of the last. The model maintains flawless character consistency, remembers what happened in previous frames, and obeys the structural logic of the environment.
The Violinist Continuity Test:
Google showcased this multi-turn consistency with a startling step-by-step evolution of a singular video asset:
“A video of a violinist playing a song.”
Prompt 2“Transport the violinist to the image environment.”
Prompt 3“Make the violin invisible.”
Prompt 4“Change the camera angle to be over the violinist’s shoulder.”
2. Deep Physical Reasoning & Complex Explainers
Omni features an upgraded intuitive understanding of real-world physics, including fluid dynamics, kinetic energy, and gravity. Instead of standard “pattern matching,” it draws on contextual meaning to manipulate elements cleanly.
The Liquid Mirror Test
“When the person touches the mirror, make the mirror ripple beautifully like liquid, and the person’s arm turns into reflective mirror material.”
The Infinite Recursive Hand Loop
“Dim the lights in the room. Put a black and white checkerboard room inside a glass sphere that floats tracking above the hand… camera slowly gets closer creating a video loop.”
The Claymation Science Explainer
“Claymation explainer of protein folding, everything is made out of clay, no hands, stop motion, accurate.”
3. “Any-to-Any” Input Combinations
Omni handles any combination of text, images, videos, and audio references to compile a single, cohesive output video. It can match a video’s motion paths, an image’s style properties, and an audio file’s beat synchronization simultaneously.
The Alphabet Marathon: Google demonstrated a rapid-fire sequence compiling 26 unique lower-third slips of paper showing items sitting on a table from a simple constraint prompt (“like a Capybara for C, disco globe for D, and Lava Lamp for L”), editing roughly 9 frames per item at 24FPS perfectly synced to background music.
Style-Shifting Walk Cycles: By referencing extreme camera distortion from one video (
video-0.mp4), a character from an image (image-0.png), and raw audio, Omni generated a front-facing full-body walking sequence that style-shifts into multiple visual genres perfectly to the beat.
4. Custom Digital Avatars
Built with responsible deployment frameworks, Gemini Omni introduces a native Avatar feature. This allows creators to safely generate a high-fidelity digital version of themselves. By using your own voice and likeness as input references, you can create videos that look and sound precisely like you for seamless, automated presentation workflows.
Access Tiers: How to Use Gemini Omni for Free
Previously locked behind closed enterprise developer tracks, Gemini Omni Flash is now live. Here is how the tier access breaks down starting today:
| Feature / Access | Free Plan Tier (Google Flow) | Paid Google AI Subscribers | YouTube Shorts / Create App |
| Daily Video Allowance | 2 Videos per day free | Unlimited / High-Priority generation | Rolling out at no cost this week |
| Editing Context | Multi-turn & One-shot | Multi-turn & One-shot | Native mobile integration |
| Content Security | SynthID Watermarked | SynthID Watermarked | SynthID Watermarked |
How to Get Started Now in Google Flow
Open up your Google Flow workspace or launch the Gemini App.
Select the new Gemini Omni Flash module canvas.
Upload your baseline assets (a photo, a drawing, an existing video clip, or a voice note).
Enter your descriptive prompt—whether a complex “one-shot” setup or a simple starting sentence—and watch your descriptions come to life.
To maintain ecosystem safety, every video exported using Gemini Omni includes an imperceptible SynthID digital watermark. These assets can easily be verified as AI-generated via Google Search, Chrome, or the Gemini App. Your two free daily clips are waiting for you right now—jump into Google Flow and start creating!
Interested in reading more about Google related AI news, rumors and deals. Read our full Google AI news coverage by clicking here.
Please follow us on our Facebook page and X account for all latest and breaking Google, Android and Nokia related news.
