Google I/O 2026 is just around the corner, and after watching The Android Show on May 12, 2026, I have a clear picture of what Google might announce at the main keynote on May 19, 2026. With consumer-centric Android announcements already out, I expect the main keynote to be a proper AI-focused event. Google may also demo new experiences related to Search, AI agents, XR glasses and developer tools. On that note, let's go ahead and learn what all we can expect at the Google I/O 2026 event.
A New Gemini Model
Google is expected to release a new Gemini model, possibly Gemini 4, at Google I/O 2026. Gemini 3.1 Pro was released in February 2026, and since then, there has been no frontier model announcement by the company so far.
That said, Alex Heath reports that the new model Google plans to announce at I/O 2026 will be roughly in the class of OpenAI's GPT-5.5, and it falls short of Anthropic's Mythos AI model. Heath also says there is "real pressure" inside Google to catch up on coding tasks in particular.
So whether the new model is called Gemini 4 or another incremental release, do not expect the generational leap that Gemini 3 was. The new Gemini model, however, will bring proactive and multi-step agentic AI.
Aluminium OS
'Aluminium OS' is Google's internal codename for the desktop operating system that likely powers the newly announced Googlebook at The Android Show 2026. Google has not officially commented on what OS it runs, but there is ample information that it's the upcoming Android-based desktop OS.
A recent leak of the Aluminum OS showed an Android-style desktop with a bottom dock, a status bar, virtual desktops, Quick Settings and more. Google has confirmed that Acer, Asus, Dell, HP and Lenovo are the first hardware partners, and the first Googlebooks will ship in fall of 2026. So I am expecting Google to officially unveil Aluminium OS at Google I/O 2026 and encourage developers to build apps for its Android desktop platform.
Android XR Glasses
Google has officially confirmed that it will preview Android XR glasses at I/O 2026. The company has been building partnerships with Samsung, XREAL, Warby Parker, Gentle Monster and Gucci. Two new Android XR glasses are expected: an audio-only AI glass similar to Ray-Ban Meta and a display one with in-lens overlays for navigation, live translation and notifications. XREAL's Project Aura may also be demoed at the stage, which is a tethered optical see-through headset.
AI Mode in Search Gets a Major Upgrade
During the Q1 earnings call, Sundar Pichai said Google is "excited to share more about Search at I/O" and put focus on Personal Intelligence, agentic booking and global Search Live availability.
Search Live has already rolled out in over 200 countries where AI Mode is available, including India. Next, Personal Intelligence, which connects AI Mode to your Gmail and Google Photos, expanded to free-tier users in the US.
So I am expecting that Google will bring Personal Intelligence globally and add agentic booking. It will allow users to book flights, hotels or restaurants in AI Mode itself. There is also the Universal Commerce Protocol (UCP) which might be demoed on the stage at Google I/O 2026.
Gemini Spark (Agent Remy)
Multiple reports say that Google staffers are already testing a major upgrade to Gemini Agent, codenamed "Remy". It may launch as Gemini Spark, based on an initial leak. It's being touted as a "24/7 personal agent for work, school, and daily life" that can take actions on your behalf and handle complex tasks proactively.
And 9to5Google has found strings in the Google app version 17.20 beta in which the upgraded Gemini agent greets with "What can I get done for you today?" It's also described as "your 24/7 digital partner" that can interact with people, share documents and make purchases using your connected apps and Personal Intelligence. Basically, Google's Project Astra vision is finally becoming real, and you may see a live demo of the feature in action.
Gemini Omni
I am most excited about Gemini Omni, which might be the new video model. According to leaks, Google is calling it a "new video model" that lets you "remix your videos, edit directly in chat, try a template, and more". Some leaked clips from Gemini Omni have appeared where a professor is working through a maths proof on a chalkboard.
The result is impressive, and Omni looks much better at prompt adherence. It's being said that Omni is an extension of Veo, and there are hints of Flash and Pro tiers. That said, it's going to be compute-heavy. One tester said that it consumed 86% of the daily AI Pro limit on just two videos. Nevertheless, expect Google to announce more about Veo, Nano Banana, Lyria 3 Pro and the Flow creative platform.
Gemini App Redesign
Google has been quietly testing a full redesign of the Gemini app and it has already leaked. The new design moves to a pill-shaped text box with a single + button that opens a bottom sheet. It houses Photos and Camera at the top and then Files, Notebooks and tools like Images, Videos, Music, Canvas, Deep Research and Guided Learning.
Some users have already got the new Gemini redesign on iOS. Google has implemented Apple's Liquid Glass design language with translucent panels and animated gradients behind the chat.
The Android version is being built on Material 3 Expressive with a blur effect rather than glass, so the two versions will feel slightly related. Note that Google has not officially announced the Gemini app redesign for Android, and it's likely we are going to see the official announcements at the I/O 2026.
Gemini Desktop Agent to Rival Claude Cowork
Google is developing a Gemini agent that can control your computer, and it's separate from the Remy agent on phones. It can use your Mac, organise your files and handle desktop tasks. By the way, Google recently launched a Gemini desktop app for macOS, so this is quite likely to happen. It looks like a direct shot at Anthropic's Claude Cowork and OpenAI's Codex agents.
Given that Google's AI models are much better at vision, the upcoming Gemini desktop agent might be better at computer use. Let's see what Google unveils at I/O 2026.
AI Ultra Lite Plan
I am also expecting that Google will introduce a new pricing tier for Gemini. Google is reportedly preparing an "AI Ultra Lite" subscription tier which will sit between the $20 Pro and $250 Ultra plans. It will offer expanded usage limits for compute-heavy tasks. Since Google is bringing new agents and a larger Omni model, a cheaper middle tier makes sense.
Wear OS 7
The Android Show on May 12 barely touched smartwatches, so I am expecting that we will hear more about Wear OS at the Google I/O 2026. The company did confirm that Gemini Intelligence is coming to Wear OS, including Create My Widget for AI-generated watch tiles. However, more details are clearly saved for the I/O event. I am expecting Google will detail Wear OS 7 with proper agentic Gemini integration, new Tile capabilities, performance gains and battery improvements.
Android 17 for Developers
There are a lot of sessions at the I/O around Android 17 "Adaptive" development. A session titled "Adaptive development for the expanding Android ecosystem" describes Android 17 as a single platform that covers phones, foldables, cars, TVs, desktops and XR through Jetpack Compose.
So expect Google to share expanded support for desktop and large-screen apps, given that Google has already unveiled the Googlebook, which runs on an Android-based desktop OS. You can also expect deep dives into performance improvements, new camera and media APIs and agentic automation hooks for developers building Gemini-powered apps.
Agentic Coding Tools
As the Google I/O 2026 is primarily a developer event, Google will share a lot of AI-powered coding tools and updates. So far, the published sessions show that Firebase is becoming an "agent-native platform" having deep integrations with AI Studio. It will also be integrated with the new Antigravity tool for vibe-coding full-stack apps.
Google is also rolling out new Gemini capabilities inside Android Studio to speed up Android development. Google is basically going after Claude Code, Codex and Cursor with its native implementation in its IDE tools. The company is expecting developers to prototype with AI Studio and ship to production on Google Cloud without leaving its stack.
Apart from that, the Gemma 4 open model, which launched in April 2026, also has a dedicated session. I am expecting new fine-tuned variants and on-device improvements rather than new open model announcements.
Google I/O 2026 is going to be packed with AI announcements, and I think we are going to see many agentic AI demos on various platforms like glasses, desktops, mobile phones, watches and more. As for Android, we have already seen some big announcements at The Android Show 2026. We will be covering the Google I/O 2026 event live, so do come back for the breakdown.
-1120w630h.webp "Gemini Intelligence on Android Explained: All Features and How to Use thumbnail")
