Google is quietly testing a new video generation model called “Omni” within its Gemini AI platform, according to early user reports and metadata analysis. The development comes as the tech giant continues to invest in AI video tools despite shifting industry dynamics, with the model showcasing early promise in handling complex visual tasks.
The “Omni” system was first flagged by a Gemini user who received a prompt to “Create with Gemini Omni,” a feature not yet officially announced by Google. The company described the tool as a “new video generation model” capable of “remixing videos, editing directly in chat, and trying templates.” While its exact relationship to Google’s existing video model, Veo, remains unclear, internal data suggests “Omni” may function as an extension of that technology.
Early demonstrations highlight the model’s ability to generate coherent, contextually rich video content. One example involved a prompt asking for “a professor writing out a mathematical proof for trigonometric identities on a traditional chalkboard, explaining the step he is currently on in the equation.” Though the output contains minor artifacts, the video successfully integrates text and movement, creating a realistic educational scene. A second test, referencing the infamous “Will Smith spaghetti” benchmark, generated a seaside dining sequence featuring two men in a high-end restaurant. The result, while not flawless, demonstrated improved handling of lighting, character interactions, and environmental details compared to earlier AI video tools.
Usage data from the test user indicates significant engagement with the feature. The two demo prompts accounted for 86% of their daily AI Pro plan usage, with some activity also recorded on Gemini Flash, a lighter-weight version of the model. This suggests “Omni” may be targeting power users or creative professionals seeking advanced video editing capabilities.
Google’s move aligns with its broader commitment to video generation, despite recent industry shifts. The company has previously emphasized its stance on the technology, stating “video’s here to stay” following OpenAI’s decision to discontinue its Sora video model in 2024. With Google I/O 2026 approaching, the search giant is expected to provide more details about its AI roadmap, including potential updates to Gemini and its video tools.
Experts note that “Omni” represents a step forward in AI video synthesis, though challenges remain. Current models still struggle with long-form coherence, physics accuracy, and nuanced human interactions. However, the ability to generate contextually relevant videos from detailed prompts marks a significant advancement in AI’s capacity to translate text into visual narratives.
As Google refines “Omni,” the tech world will be watching to see how it differentiates from competitors. With major industry events on the horizon, the coming months could reveal whether this new tool signals a broader shift in how AI shapes digital content creation.
For now, the “Omni” project remains in its early stages, but its early demos suggest Google is continuing to push the boundaries of what AI can achieve in video generation.
