From Aluminium OS to Gemini Omni: The biggest AI upgrades expected at Google I/O 2026

Google I/O is all set to kick off from tomorrow where the company will showcase some of its biggest AI upgrades of the year. Just like last year, Google held the Android Show separately where it showed the latest features coming to Android 17 including Gemini Intelligence, Google Chrome's AI integration, Rambler AI tool inside Gboard and Geminibooks.

While Google has continued to stay mum on the new AI fetures it could introduce at I/O 2026, various leaks have given us a good idea of what to expect from the tech giant's one of the biggest events of the year.

Top upgrades expected from I/O 2026:

1) Gemini 4.0:

Google last launched its Gemini 3.1 Pro model back in February 2026 and its about time for the company to get back in the AI chatbot race which has seen various new entrants from GPT-5.5 to Claude Opus 4.7 and the upcoming Mythos model Anthropic.

It's not entirely clear what the new AI model could be named but a new model launch is more or less par for the course at this point. Google had also launched the Gemini 2.5 Pro model at I/O last year.

Google has also been tipped to be blind testing various AI models dubbed Ajax, Hercules, Hector, and Orpheus which are likey the internal names for the AI models before they are released to the public.

2) Gemini Spark:

As per a report by 9to5Google, the tech giant is looking to add more advanced agentic AI capabilities to the Gemini app. The new capabilities could be branded as ‘Gemini Spark’ and could allow the chatbot to complete tasks on behalf of the users.

With the upgrade, Gemini could reportedly automate tasks like decluttering inboxes by summarising or archiving newsletters, preparing meeting briefs, and creating personalised news digests that track stories over time.

3) Gemini Omni:

There have also been leaks about Google testing a new video generation model called ‘Gemini Omni’. The new model was spotted after a user on Reddit posted a screenshot of Gemini Omni being available to them.

The description for new model read, “Meet our new video generation model. Remix your videos, edit directly in chat, try a template, and more.”

4) Gemini Live:

A report by Forbes found that Google is internally tesiting a number of new AI models for Gemini AI live which it notes were disvoered via a hidden selector inside the Google app.

The report notes that there are measurable differences between these models with some models being able to access live weather and location while others having stronger memory and context awareness.

One model, codenmaed, ‘Capybara’ also identifies itself as ‘Gemini 3.1 Pro’ instead of the Flash model that Google typically uses.

5) Aluminium OS:

Google already unveiled its Geminibooks at the Android show and it's now time for the company to reveal the operating system running it.

Aluminum OS has been in the works for a while and is expected to be the desktop focused OS which essentially brings Android to bigger screens.

Ahead of the Android Show, a 16 minute leaked video had shown off the Aluminium OS with various l desktop-style features, including desktop folders, virtual desktops, an optimised Quick Settings panel along with redesigned notification interface, and tighter ecosystem connectivity

The UI also included a bottom app dock, app drawer, task-manager style tools, and workspace switching through virtual desktops.