Google's Mysterious 'Omni' Video Model: What the Gemini UI Leak Tells Us Ahead of I/O 2026

Google’s Mysterious ‘Omni’ Video Model: What the Gemini UI Leak Tells Us Ahead of I/O 2026

A small UI string is the entire story so far, but it’s a string with very interesting placement. TestingCatalog reports that a fresh screenshot from Gemini’s video generation tab carries a previously-unseen line: “Start with an idea or try a template. Powered by Omni.” Omni shows up right next to “Toucan” — the internal name for Gemini’s current video tool, the one already powered by Veo 3.1.

Two weeks before Google I/O 2026 opens (May 19–20), that’s a hard rumor to ignore. Here’s what the leak suggests, what’s still ambiguous, and where this fits into the larger video-generation race in 2026.

What was actually spotted

The single piece of evidence is a UI placeholder text from Gemini’s video generation experience. Two details make it more than noise:

The string is visible to users, not just buried in source code or feature flags. UI copy that mentions a brand name typically reaches that state only when the team is preparing for a public release.
The placement is next to “Toucan” — a known internal codename for Google’s current Veo-3.1-backed video generation pathway in Gemini. New code lives next to old code. A new product name parked next to an existing one is the standard staging pattern before a swap.

TestingCatalog’s Threads post summarises it: “If Google plans to release Gemini Omni for video generation, it would likely outperform Veo 3.1.”

What “Omni” could actually be

Three readings, in increasing order of how transformative they’d be:

1. A new Veo wrapper

The least disruptive explanation: Omni is the new product name for Gemini’s video tab, with Veo 3.x or Veo 4 still doing the actual generation work. Brand consolidation under a single Gemini-native name (the way Nano Banana sits on Gemini 3 / 3.1 Flash Image for stills) would explain why a public-facing string is appearing at all.

2. A new Gemini-trained video model

Google may have trained an in-house video model under the Gemini umbrella to sit alongside or replace Veo. The split today — Veo for video, Nano Banana / Gemini for images — is awkward enough that the model team has reasons to unify it. Omni would be the result, with its own architecture and benchmark profile distinct from Veo 3.1.

3. A true omni-model — single system for image + video + more

The most ambitious read, and the one the name suggests: a single Gemini omni-model that handles image generation, video generation, and possibly audio in the same system, the way GPT-4o is positioned for text-image-audio. If true, Gemini would be the first top-tier omni-model with video output — a meaningful first.

The leaked string can’t distinguish between these. But option (3) is the only one that justifies a brand-new public name like “Omni” rather than just bumping Veo’s version number.

Why this matters for the video-gen race in 2026

Video generation is the most competitive category in generative AI right now. The current state of the leaderboard, roughly:

ByteDance Seedance 2.0 is at the top of public benchmarks, with the Fast and Turbo variants making cinematic AI video financially viable for high-volume production.
Alibaba Wan 2.7 ships text-to-video, image-to-video, video-edit, and reference-to-video with audio-synced motion at 1080p.
Kuaishou Kling V3.0 (Std + Pro + O3) anchors the Asia-led wave alongside Wan and Seedance.
OpenAI Sora 2 is widely available now, with a Pro variant for higher resolution.
Google Veo 3.1 holds the high ground for cinematic camera work and audio-visual sync, but has been gated and region-locked.

A Gemini Omni release that genuinely outperforms Veo 3.1 — the bar TestingCatalog sets — would be the first top-tier multi-modal unified model with video output, and the first time Google has chosen to compete on a single-name product line rather than a Veo / Imagen / Nano Banana split.

When to expect the reveal

Google I/O 2026 runs May 19–20, 2026. Gemini and AI updates are confirmed agenda items. A pattern of pre-I/O UI leaks surfacing a fresh public name is consistent with a keynote-stage reveal.

That said, treat all of this as speculative until Google says it on stage. UI strings have shipped without product launches before. The most defensible read of the leak is: Google has a video product called Omni in late-stage staging on Gemini, and the most plausible window for a launch is the next two weeks.

What this would mean for WaveSpeedAI users

Veo 3.1 is already available on WaveSpeedAI, and we expect to add any new Gemini video model to the platform shortly after public release — same single API key, same OpenAI-compatible LLM endpoint pattern, same per-second pricing.

If Omni replaces or supplements Veo 3.1, you’ll be able to A/B-test it against the rest of the video-gen lineup we already host:

Seedance 2.0 — current SOTA on benchmarks
Wan 2.7 — Alibaba’s cinematic video model
Kling V3.0 Pro — Kuaishou’s high-fidelity option
Sora 2 — OpenAI’s offering
Veo 3.1 — current Google video model

If Google ships Omni at I/O, expect to compare it under the same API in the days after launch.

Frequently asked questions

What is Google’s “Omni” video model?

Based on a UI string spotted in Gemini’s video generation tab, Omni appears to be a new Google video-generation model or product brand that replaces or supplements the current Veo-3.1-powered Gemini video flow. Nothing official has been announced.

Will Omni replace Veo 3.1?

Unclear. Three plausible interpretations: (1) Omni is a new public name for the same Veo-powered pathway; (2) Omni is a new Gemini-trained video model alongside Veo; (3) Omni is a unified Gemini omni-model handling image and video in one system. Only Google can confirm which.

When will Google reveal Omni?

The most likely window is Google I/O 2026 on May 19–20, where Google has confirmed Gemini and AI updates are on the agenda.

Will Omni outperform Seedance 2.0?

ByteDance Seedance 2.0 currently tops most public video-gen benchmarks. Whether Omni can leapfrog it depends entirely on what’s actually under the hood — none of which is known yet.

Will Omni be available on WaveSpeedAI?

If Google releases Omni publicly, it will be added to WaveSpeedAI’s model catalog shortly afterward, alongside the existing Veo 3.1, Seedance 2.0, Wan 2.7, Kling V3.0, and Sora 2 video models.

Watch this space

Two weeks until I/O. We’ll update this post the day Google confirms anything; in the meantime, the existing Veo 3.1 and the rest of the video-gen lineup are live on WaveSpeedAI under one API.

Source: TestingCatalog (web post, Threads).