New Google Gemini 3.2 + Omni LEAKS!

New Google Gemini 3.2 Plus Omni Leaks New Google Gemini 3.2 Plus Omni just leaked, and this one is bigger than people think. Two weeks before Google I.O. 2026, a screenshot showed up inside the Gemini app. One line of text.

Just one. It said, Start with an idea or try a template. Powered by Omni. That's it.

That's the whole leak. But that one line tells us Google is about to ship something we have not seen before. So here's what's going on. Right now, when you open Gemini and try to make a video, it says, Powered by VO 3.1.

That's their video model. For images, it's Nano Banana 2 and Nano Banana Pro. Different model for each job. Different name for each tool.

That's how Google has been doing it for the past year. But this leak shows something new. The line, Powered by Omni, replaced the VO 3.1 tag. Same spot.

Same screen. New name. And the placement matters because that text is in the live app, not buried in code. When a brand name shows up in the actual user interface, the team is days away from shipping.

Not months. Days. The leak came from Testing Catalog first. Then it got picked up by Wes Roth, Wavespeed AI and a bunch of others.

Google I O 2026 runs May 19th and 20th at Shoreline Amphitheatre. So we are looking at a launch in two weeks. Maybe sooner. Now here is the part that matters.

Omni is not just a new video tool. The word Omni means everything. One model. One brand.

Video, images, maybe even audio, all in the same system. Right now, Google splits this work across three or four models. Omni puts it all in one place. Like how GPT 4.0 handles text, image and audio together.

Google has never done that with video before. If this is real, Gemini becomes the first top-tier AI that can do high-end video and images inside one unified system. And there is more. A second leak came out at the same time.

This one is the iOS app redesign. They are calling it Liquid Glass. The whole app got a fresh look that matches iOS 26's design. It looks clean.

Clear. See-through panels. New keyboard. New menus.

Reddit users on our Gemini AI started posting screenshots. People got really excited. But the redesign is not the real story. The real story is what the new menu is for.

The Liquid Glass menu is built to house something called computer use tools. That means Gemini is about to get the power to see your screen and act on it. Not just talk to you. Act.

Click stuff. Type stuff. Open apps. Move between tabs.

All on its own. This connects back to a project Google has been working on for a long time. It's called Project Jarvis. Some leaks call the new system Project Jarvis 2.

The idea is simple. You tell Gemini what you want done. It opens Chrome. It looks at the screen with its own AI vision.

It clicks. Types. Fills forms. And finishes the job.

You go grab a coffee. It books your flight. It fills out the form. It pulls research from 20 tabs.

Done. Real quick before we keep going. If you want to actually use Gemini Omni to grow your business the second it drops, come join us inside the AI Profit boardroom. We have four coaching calls every week where we go deep on how to plug new Google AI tools into your day-to-day work.

The moment Omni goes live, we are walking you through exactly how to use it for content, lead gen, and client work step-by-step. Plus a full 30-day roadmap for using Gemini agents in your business, daily tutorials, and a member map so you can connect with other Gemini users near you. Link in the description or go to aiprofitboardroom.com Okay, back to the leaks. Let's break down what Omni can actually do based on the leaked details.

There are a few big ones. First, autonomous web navigation. The leaked internal docs mention that Omni runs on the new Gemini 3.1 Pro architecture. It can open Chrome and book a flight, run deep research, or fill out a form without you touching the keyboard.

You give it the goal. It does the steps. Second, the universal context engine. This part is interesting.

The leak says Gemini Omni can hold persistent intent across many browser tabs and apps at the same time. Most AI tools forget what you were doing when you switched tabs. Omni does not. It knows what you opened in tab 1, what you typed in tab 3, and what you bought in tab 7.

It keeps the thread. So when you ask a follow-up question two hours later, it still knows what you meant. Third, the vision action loop. This is the big shift.

Omni does not just read the page like a normal browser tool. It looks at raw pixels. It takes screenshots. It figures out where buttons are.

It generates clicks just like a person would. This means it works on any website, even old ones, even ones that were never built for AI. If a human can use the site, Omni can use the site. Fourth, PhD level reasoning with low latency.

The leak says Omni will offer deep reasoning at near zero delay. So it thinks fast and thinks well. That combo is rare. Most smart models are slow.

Most fast models are dumb. Omni is supposed to do both. Now let's talk about Gemini 3.2 itself. Because Omni and 3.2 are linked.

A leaked post on R Singularity said Google plans to unveil Gemini Omni alongside Gemini 3.2 and 3.5 at IO. The post is unverified, but it lines up with the pattern. Google released Gemini 3 Pro in November 2025. Then 3.1 Pro in March 2026.

The cadence is fast. Every few months. So a 3.2 in May fits the timeline. Gemini 3.1 Pro already scored 77.1% on the R-Cage IT benchmark.

That test measures how well an AI can solve brand new logic puzzles. It's more than double what came before. And 3.1 Pro is already inside the Gemini app, Notebook LM, Vertex AI, Antigravity and Android Studio for paid users. So 3.2 will likely push this even higher.

Here's another thing that matters. Google Cloud CEO Thomas Kurian recently hinted that a new Gemini version is arriving very, very soon. That came in a public discussion about AI infrastructure. When the cloud CEO drops a hint like that two weeks before IO, it's not random.

It's a signal. So what does all this mean for someone running a business? Let me make it real simple. Right now, doing tasks online takes you time.

You open tabs. You click. You read. You compare.

You fill forms. You wait. You repeat. Most days, half your work is just clicking and reading.

That's where the time goes. Omni changes that. You give it the goal. It does the clicks.

You skip ahead to the part where the work is done. And here's why I think this is bigger than people are saying. Google has the browser. They own Chrome.

Most people on the internet use Chrome every day. So when Google ships an agent that lives inside Chrome, they are not adding a new tool. They are upgrading the tool you already use. That's a different kind of rollout.

That hits everyone fast. OpenAI has Sora too. ByteDance has Cdance 2.0. Alibaba has One 2.7.

They all make great video. But none of them sit inside the browser most people use. None of them have the Chrome tab. None of them have Gmail, Calendar, Drive and Docs already wired in.

Google has all of that. So when Omni ships, it does not need to win a new market. It just needs to plug into the market Google already owns. Now let me cool things down a bit.

There is real stuff we don't know yet. The leak is one UI string. That's all. Google has not confirmed any of this.

UI strings have shipped without products before. So treat this as very likely, not 100% done deal. We will know for sure on May 19th. Also the Gemini 3.2 and 3.5 part is the weakest piece.

That came from a Reddit post. It might be true. It might be guesswork. The Omni leak is much more solid because it shows up in the real app.

There is also the question of trust. An AI that can click around in your browser is helpful. But it can also do things you didn't mean to. The EU AI Act now says autonomous agents must keep logs and have a kill switch.

Google will have to build that in. Otherwise people won't use it for anything important. But even with all that, the direction is clear. AI is moving from answer my question to do my work.

And Gemini Omni is one of the biggest steps yet. Quick story. There is a company called Wavespeed AI that runs every video model under one API. They cover VO 3.1, SORA 2, CDANCE 2.0, WAN 2.7 and more.

Their team wrote that if Omni really is a unified model, it would be the first top tier video tool inside an Omni model. They are not hyping it. They are building on it. When the people who run the backend say this, it carries weight.

So what do you actually do right now? While we wait for IO on May 19th and 20th. Step one. Open the Gemini app.

If you have iOS, watch for the liquid glass update. Check your settings for new menu options. Some people are getting it early. Step two.

Start playing with Gemini 3.1 Pro right now. It's already in the Gemini app for AI Pro users. It's in Notebook LM. It's in Google AI Studio if you want to test the API.

Get used to how it thinks because 3.2 will feel similar but smarter. Step three. If you have not tried VO 3.1 yet, do it. Make a few short videos.

See how it handles motion, audio and prompts. When Omni replaces it, you will already know the basics. Step four. Try the computer use tool that's already in Gemini 3 Pro.

That feature is live. It can already control a webpage, click buttons and fill out simple forms. Omni is a better version of that. So learning the current one gives you a head start.

Step five. Watch the Google IO keynote on May 19th. It runs at Shoreline Amphitheatre. Sundar Pichai will likely lead.

The keynote is when they confirm or deny everything we just talked about. If you want to be ready for Gemini Omni the day it launches, come join us inside the AI Profit boardroom, building real workflows with the latest Google tools. Four weekly coaching calls focused on Gemini, VO, Notebook LM and the new agent tools as they drop. A 30-day roadmap built around Gemini agents so you know exactly what to do week by week.

Daily tutorials with step-by-step videos. A prompt library packed with Gemini prompts that work. And a member map so you can find and message other Gemini users near you for help anytime, day or night. Link in the description or go to aiprofitboardroom.com.

And if you want the full process, all the SOPs and over 100 AI use cases like this one, come join the AI Success Lab. It's our free community with 67,000 members who are figuring out AI together every day. You'll get all the notes from this video plus access to the full community. Links in the comments and description.