Google Accidentally Leaks "Gemini Omni" Days Before I/O, Hinting at a Unified Video-Image Model

Six days before Google I/O 2026, the name Gemini Omni showed up inside the Gemini app. This wasn't a stray code string buried in a developer build. Users saw it, generated videos with it, and posted the receipts online. Google may be about to reshape how it delivers creative AI.

What the Gemini Omni leak actually tells us

A Reddit user gained early access to a "Gemini Omni" model inside the Gemini app, described as Google's new video generation tool with chat-based editing and templates. The first hint appeared on May 2 when an X user spotted the label "Powered by Omni" in Gemini's video tab. TestingCatalog picked it up, and the AI community ran with it.

Why Gemini Omni could be Google's biggest creative AI bet

Google currently runs a fragmented setup: Veo 3.1 handles video, Nano Banana manages images. Omni could merge both under one Gemini-native brand. If true, it would be the first top-tier model that generates both images and video in a single system.

The timing is almost too perfect. OpenAI shut down Sora on April 26, 2026, leaving a gaping hole in consumer video generation. ByteDance's Seedance 2.0 tops most video benchmarks right now, Runway Gen-4 and Kling 3.0 remain strong, but none of them handle text, images, and video under one roof.

Early outputs show promise and familiar limitations

A math-chalkboard test video impressed with accurate reasoning and lifelike rendering. A dinner-scene prompt was less convincing, with objects materializing from thin air. The model currently appears limited to 10-second clips, and two video generations reportedly consumed most of a daily AI Pro quota.

Google I/O 2026 and the Gemini Omni reveal

Google I/O runs May 19–20 with Gemini and AI updates confirmed on the keynote agenda. Parallel leaks reference Gemini 3.2, Gemini 3.5, and a visual model codenamed "Spark Robin." Omni will likely ship in Flash and Pro tiers, following the pattern Google used for earlier Gemini model launches.

For content creators, the value proposition is straightforward: storyboard, still images, and video output from a single chat window. No more export-import-rerender chains across three different tools. Even if Omni's raw quality trails Seedance 2.0 at launch, that workflow compression alone could make it the default creative tool for millions of Gemini users.