Google appears to be preparing a new video generation model called Gemini Omni, based on leaked UI strings, mobile app references, and early user reports surfacing ahead of Google I/O 2026 on May 19-20. The company has not officially announced the model, but multiple signals suggest it could introduce conversational video editing, template-based creation, and notably better handling of text and equations in generated video. Whether Omni represents a rebrand of the existing Veo model, a new Gemini-native system, or a true unified model that handles text, images, and video in one system remains unclear.

What Exactly Has Leaked About Gemini Omni?

Three waves of evidence have emerged in recent weeks. First, a user-visible string appeared in Gemini's video generation tab reading: "Start with an idea or try a template. Powered by Omni." This placement next to "Toucan," the internal codename for Gemini's current Veo 3.1-powered video pipeline, follows the standard staging pattern before a product swap.

Second, Reddit users and mobile app testers discovered additional references inside the Gemini app, including the description: "Meet our new video model. Remix your videos, edit directly in chat, try a template, and more." Early impressions from these testers highlighted strong prompt adherence, smooth camera angle transitions, improved scene coherence, and notably better voice generation quality. One particularly striking demo showed a professor writing math equations on a blackboard, with the equations reportedly rendering correctly in the generated output, a feat that requires both visual coherence and semantic accuracy.

Third, demo coverage has broadened beyond niche leak tracking into mainstream AI media discussion. The most repeated demo themes include the math equation example, one-sentence video edits, object replacement, stylized animation output, and fast credit consumption during early testing. A model ID discovered in the app, "bard_eac_video_generation_omni," also revealed a 10-second generation limit, which is short by current standards.

How Would Gemini Omni Change Video Creation Workflows?

If the reported features ship as described, they would represent a significant shift in how creators interact with AI video tools. The most transformative capability appears to be chat-based editing, which would turn Gemini into a conversational video editor rather than a traditional prompt-and-wait system. Instead of generating a video and starting over if changes are needed, creators could describe edits in natural language directly within the chat interface.

Additional reported features suggest a move toward post-production control and creative iteration:

• Video Remixing: The ability to remix existing videos rather than always starting from scratch, moving beyond pure text-to-video generation into edit-and-remix workflows.
• Templates: Pre-built starting points aimed at mainstream creators, lowering the prompt engineering barrier but potentially driving output homogeneity.
• Object Replacement: Swapping elements within generated videos for creative iteration, though brands would need provenance, consent, and rights guardrails before relying on remix workflows.
• Text Coherence: Improved handling of text and equations in video, a genuinely difficult technical challenge that signals meaningful improvement if reports hold.

Is Omni a New Model or Just a Rebrand?

The AI video community is debating three plausible interpretations. The first scenario is that Omni is simply a rebrand of Veo for consumers. Google could retire the Veo brand in consumer-facing products and replace it with "Omni" as a unified identity, similar to how image generation was consolidated under the Nano Banana name. The underlying model might still be Veo 3.x or Veo 4. This interpretation is moderately likely, as brand consolidation is a plausible reason for a new name.

The second scenario is that Omni is a new Gemini-native video model, a version of the Gemini architecture fine-tuned specifically for video output and architecturally separate from the Veo model family. This would mean Google is running two parallel video model tracks: Veo for API and enterprise use, Omni for Gemini consumer experiences. Google has done this before with its image models, making this scenario moderately likely as well.

The third and most ambitious interpretation is that Omni is a true omni-model, a single Gemini model that natively generates text, images, video, and potentially audio within one unified system. This would make Gemini the first major omni-model with native video output, a meaningful first in the space. However, this scenario is considered lower likelihood, though the name "Omni" explicitly suggests it.

What Should Creators Do Right Now?

Google has not officially announced Gemini Omni, and no one outside the company has confirmed access to a stable, public-facing version. Until Google publishes a product page, model card, API route, pricing, and usage limits, all demo claims should be treated as pre-announcement signals rather than confirmed features.

For production work today, creators should rely on live tools such as Veo 3.1 and PixVerse V6 rather than waiting for an unconfirmed model. The next official watch window is Google I/O 2026 on May 19-20, where Google is expected to make announcements. Developers should not plan around Omni API availability until Google formally confirms it, as API access has not been confirmed in any of the leaked reports.

The broader significance of these leaks is that they signal Google's direction in video generation: toward conversational interfaces, template-based workflows, and improved handling of complex visual elements like text and equations. Whether Omni ships as described, gets delayed, or transforms into something different entirely, the leaked features offer a glimpse into how major AI companies are thinking about the next generation of creative tools.