Google Unveils Gemini Omni Flash: AI That Creates Videos from Any Input
Google has officially launched Gemini Omni Flash, a groundbreaking new AI model that can generate and edit videos from virtually any combination of inputs — including text, images, audio, and even other videos. Announced on the official Google Blog, the model represents a significant leap forward in multimodal AI capabilities, bringing together Gemini's world-class reasoning with advanced video generation.
Gemini Omni Flash is the first model in the new Omni family, and it is rolling out now to the Gemini app, Google Flow, and YouTube Shorts. The model allows users to create high-quality videos grounded in Gemini's real-world knowledge, all through simple natural language instructions.
Create Anything from Any Input
What sets Gemini Omni apart is its ability to accept multiple input types simultaneously. Users can combine images, audio clips, video references, and text prompts into a single cohesive output. For example, a user could provide a photo of a character, a video showing a specific camera movement, and an audio track for background music — and Omni will blend them all into a seamless video clip.
The model also excels at video editing through conversation. Users can make iterative changes using natural language, with each instruction building on the last. Characters remain consistent, physics holds up, and the scene remembers what came before — enabling complex multi-turn editing workflows that were previously impossible with consumer AI tools.
Intelligent Video Creation Grounded in Real-World Knowledge
Unlike traditional AI video generators that rely purely on pattern matching, Gemini Omni draws on Gemini's deep understanding of physics, history, science, and cultural context. This allows it to create scenes that not only look realistic but also make logical sense. The model demonstrates an improved intuitive understanding of forces like gravity, kinetic energy, and fluid dynamics, resulting in more physically accurate video outputs.
Google demonstrated the model's capabilities with examples ranging from transforming a sculpture into bubbles, to creating a claymation-style explainer of protein folding, to generating a rapid-fire alphabet video with 26 unique items — all from simple text prompts.
Responsible AI with Built-in Safeguards
Google has implemented several safety measures for Gemini Omni. All videos created with the model include an imperceptible digital watermark (SynthID), allowing users to verify AI-generated content through the Gemini app, Gemini in Chrome, and Google Search. The company has also established clear policies governing the use of its AI tools, with additional features like digital avatar creation being rolled out carefully.
Initially, users can create videos featuring their own digital avatar using voice references. Google stated it is continuing to test and understand how to responsibly bring audio and speech editing capabilities to users.
Availability
Gemini Omni Flash is rolling out today to all Gemini app users and is available at no cost to users on Google Flow. In the coming weeks, Google plans to make the model available to developers and enterprise customers via APIs. The company also indicated that future models in the Omni family will support additional output modalities including image and audio generation.
This launch marks a significant milestone in the evolution of generative AI, moving beyond simple text and image generation into a truly multimodal creative assistant that can understand, reason about, and manipulate video content through natural conversation.
Image Source: Google
