Gemini Omni May 2026 Release Notes: What Shipped at I/O and What's Coming Next

TL;DR — what is actually live on May 27, 2026

• Gemini Omni Flash went live globally on May 19, 2026 during Google I/O.
• Available free on YouTube Shorts Remix and YouTube Create for users 18+.
• Available in the Gemini app and Google Flow for Google AI Plus ($7.99/mo), Pro ($19.99/mo) and Ultra subscribers.
• Output is video, up to 10 seconds per clip, with native audio and SynthID watermarking on every generation.
• The developer / enterprise API for Omni Flash is not live yet — Google says “in the coming weeks” via the Gemini API and Vertex AI.

If you only want to remember three things from I/O 2026: Omni Flash is shipped, the API isn’t, and Omni Pro is next.

The launch surface, by product

Gemini app

Omni Flash replaces the previous Veo 3.1-backed video generator inside the Gemini app for Plus / Pro / Ultra subscribers. The big shift is conversational editing: once a clip is generated, you can ask for “make the background a rainy Tokyo street” or “give him a leather jacket” without re-prompting from scratch. The model keeps full context across turns and preserves character identity.

Google Flow

For creative pros, Flow is now the most powerful Omni surface. You can blend reference images, audio clips and existing footage into a single Omni prompt, then iterate scene-by-scene while keeping a consistent character across shots. The chat-driven scene timeline is the headline UX change of the May 19 release.

YouTube Shorts Remix

This is the first time a frontier video model has landed natively on a 2-billion-user platform on launch day. Pick an eligible Short, prompt the change (drop yourself into the scene, add a new visual reference, swap the setting), and publish — all without leaving the YouTube mobile app. Every Omni clip is auto-labeled in Shorts discovery as AI-generated through SynthID.

YouTube Create

The desktop / tablet creator suite gets the same Omni Flash access as Shorts Remix, with longer-form chaining inside a project file. The cap is still 10 seconds per individual generation.

What’s new in the model itself

Compared with the leaked Veo 3.1 era, the launch model adds:

• Physics-aware rendering — fluids, cloth and object interactions hold up plausibly across edits.
• Character consistency — identities and voices stay stable across multiple shots and turns.
• AI Avatars — a personal digital likeness you set up once and reuse in future videos. Voice references only at launch; broader audio reference types coming.
• Conversational, multi-turn editing — full context retained turn-to-turn; no re-prompt from scratch.
• SynthID watermark on every clip — verifiable through the Gemini app, Gemini in Chrome and Google Search.
• A single unified stack — Omni collapses what used to be Veo (video) + Imagen (image) + separate audio systems into one model, which should reduce cross-modality artifacts.

Google has not published numeric benchmarks alongside the launch, so independent eval is still pending — the most consequential follow-up will come when the API opens up.

Pricing changes that landed alongside Omni

I/O 2026 also restructured Google’s AI subscription tiers:

Tier	New monthly price	What you get
AI Plus	$7.99	Omni Flash in Gemini app + Flow, entry-level limits
AI Pro	$19.99	Higher Omni limits, more Flow credits
AI Ultra (entry)	$100 (new tier)	5× higher Gemini & Antigravity limits, 20TB cloud storage
AI Ultra (full)	$200 (was $249.99)	Highest limits, priority access, everything

The new $100 AI Ultra tier is specifically pitched at developers, technical leads and advanced creators — it’s the tier most early Omni power users will land on once the API opens.

What’s still coming — the public roadmap

Google was explicit about what is not in the May 19 release but is on the way:

1. Developer & enterprise API — Omni Flash via the Gemini API and Vertex AI, “in the coming weeks.” No firm date, no official pricing. Analyst estimates put it in the $0.10–$0.30 per second of video range, which would land it competitively against Runway Gen-4 and below Veo for many workloads.
2. Gemini Omni Pro — the more capable model in the Omni family, teased for the coming months. Expected to extend clip length, raise resolution, and likely land first on the new $100 AI Ultra tier.
3. More audio inputs — voice references are supported at launch; richer audio input types and audio output modality are explicitly on the roadmap.
4. Image output modality — Omni starts video-first, but the long-term promise is “any input → any output.” Image and audio outputs are publicly committed to but not dated.
5. Wider geo + Workspace integration — staged rollout continues; if you don’t see Omni entry points in your account yet, that’s expected.

What this means in practice

• If you publish on YouTube, you can start using Omni today, for free, with no waitlist.
• If you build a SaaS or production pipeline on top of Omni, hold for the API — there is no supported way to integrate Omni Flash into a backend until the Gemini API / Vertex AI rollout lands.
• If you’re picking a subscription tier, the new $100 AI Ultra is the inflection point for serious Omni use, especially once the API ships.
• If you’re a Veo 3.1 user inside the Gemini app, you’re effectively on Omni now — re-test your prompts for character consistency and conversational editing wins.

We’ll keep this post updated as the API timeline firms up and as Omni Pro details are announced.