Google DeepMind just unveiled Gemini Omni, a new AI model that can accept multiple types of input (text, audio, images, and video) and generate realistic, physics-accurate videos in response. The first version, called Gemini Omni Flash, rolled out on May 20, 2026, to paid subscribers of Google AI Plus, Pro, and Ultra, with a free launch coming to YouTube Shorts and YouTube Create later that week.

What Makes Gemini Omni Different From Other AI Video Tools?

Unlike existing text-to-video tools such as Google Veo, Gemini Omni is multimodal in both what it accepts as input and what it produces as output. This means users can feed the model text, audio, images, or video, and Omni will generate a unique, interactive world using Gemini's real-world knowledge. The model can produce videos with more accurate physics simulations, resulting in more realistic-looking content that understands the context of a prompt, such as a historical fact, to generate more accurate video content.

Beyond simple video generation, Gemini Omni enables conversational video editing. Users can take a video they shot or an AI-generated video and edit specific aspects through natural language commands. If a user likes a shot but wants to change the background, they can do so with Omni. The model can modify the style, angle, scenery, or even a specific detail in a clip.

How to Use Gemini Omni Flash for Video Creation and Editing

• Generate Videos From Multiple Inputs: Feed the model text descriptions, audio clips, images, or existing video footage, and Omni will create a new video that combines your inputs with realistic physics and contextual accuracy.
• Edit Videos Through Conversation: Instead of using traditional editing software, describe the changes you want in natural language, such as "change the background to a forest" or "make the lighting warmer," and Omni will apply those edits.
• Create Digital Avatars: Users can generate their own digital likeness through Avatars, though Google is still testing this feature to ensure responsible deployment before full rollout.
• Verify AI-Generated Content: All videos created with Omni are embedded with the SynthID watermark, allowing viewers to verify that the content was AI-generated rather than authentic footage.

Why Is Google Calling This a Step Toward AGI?

During the presentation at Google I/O 2026, Google DeepMind CEO Demis Hassabis described Omni as a crucial milestone on the path to artificial general intelligence, or AGI, which refers to AI systems that can perform any intellectual task that a human can. Hassabis stated that in the future, Omni would be able to output "anything" the user wanted, positioning the model as a foundational step toward more capable AI systems. The multimodal nature of Gemini Omni, combined with its ability to understand real-world context and generate physically plausible content, represents a shift from narrow, single-purpose AI tools to more flexible, general-purpose systems.

"Gemini Omni is a new model that can create anything from any output," said Demis Hassabis, describing the model as a crucial step toward AGI.

Demis Hassabis, CEO of Google DeepMind

The emphasis on world models, which simulate how the physical world behaves, reflects a broader industry belief that understanding and predicting real-world dynamics is essential for building more intelligent AI systems. Gemini Omni's ability to generate videos with accurate physics suggests that Google DeepMind is making progress on this front.

When and Where Can You Access Gemini Omni Flash?

Gemini Omni Flash became available on May 20, 2026, to paid subscribers of Google AI Plus, Pro, and Ultra within the Gemini app and Google Flow. Later that same week, the model launched in YouTube Shorts and the YouTube Create app at no cost to users, making the technology accessible to a broader audience without requiring a paid subscription. This tiered rollout strategy allows Google to gather feedback from power users first before expanding to the general YouTube creator community.

The availability across multiple platforms signals Google's confidence in the model's safety and quality. The inclusion of the SynthID watermark on all generated videos addresses growing concerns about AI-generated content being mistaken for authentic footage, a critical consideration as video generation technology becomes more realistic.

What Does This Mean for Content Creators and AI Development?

Gemini Omni Flash represents a significant expansion of what AI can do in the creative space. Rather than replacing human creators, the tool offers a new way to iterate on ideas, edit footage, and explore creative directions without learning complex software interfaces. The conversational editing feature, in particular, lowers the barrier to entry for video production, potentially democratizing a skill that traditionally required specialized training and expensive equipment.

For the broader AI industry, Gemini Omni demonstrates that multimodal models capable of understanding and generating multiple types of content are becoming practical reality. The model's emphasis on physical accuracy and contextual understanding suggests that Google DeepMind is moving beyond simple pattern matching toward systems that reason about how the world works. This aligns with the industry consensus that world models will be central to the next generation of AI capabilities.