Google I/O 2026 was not a subtle event.
Google came in with a message that was hard to miss: we are no longer behind. From new AI models to a complete redesign of Search, an autonomous personal agent, an AI shopping cart, and glasses made with Gentle Monster — it was a lot to process in one sitting.
So we did the sitting for you.
Here’s everything that matters from Google I/O 2026, broken down clearly, without the fluff.
Quick Answer: What Did Google Announce at I/O 2026?
Google announced Gemini 3.5 Flash, a new frontier AI model that outperforms earlier pro models at faster speeds. They launched Gemini Omni, a video-first multimodal model that can turn any input — text, image, video, audio — into a single output. They unveiled Google Antigravity, an agent-first development platform replacing their earlier tools. They redesigned Google Search with AI at its core, introduced a Universal Cart for cross-platform shopping, launched a personal AI agent called Gemini Spark, and introduced audio/display glasses built in partnership with Gentle Monster, Warby Parker, and Samsung.
That’s the short version. The longer version is below.
Part 1: The New Models
Gemini 3.5 Flash — Speed Without Sacrifice
The headline model from this year’s I/O is Gemini 3.5 Flash.
What makes it interesting is not just what it can do — it’s the tradeoff it refuses to make. Most AI models make you pick between speed and quality. Gemini 3.5 Flash sits in what Google calls the “top-right quadrant” of the Artificial Analysis index — meaning it delivers frontier-level intelligence and is fast enough for real-time use.
Numbers: it outperforms Gemini 3.1 Pro on benchmark tests for coding and agentic tasks — Terminal-Bench 2.1 (76.2%), GDPval-AA (1656 Elo), and MCP Atlas (83.6%). It’s generally available today through the Gemini API, Google AI Studio, and Android Studio.
Where it shines practically: long-running tasks that used to take developers days or auditors weeks. The model can plan, build, and iterate across complex problems — coding new apps, maintaining codebases, preparing financial documents.
Gemini 3.5 Pro is coming next month. It’s already being used internally at Google.
Gemini Omni — Any Input, Any Output
Gemini Omni is the more ambitious announcement.
The idea: one model that can generate anything from anything. You give it text, images, video, audio — it understands all of it and creates a cohesive output. Right now, it starts with video generation. Over time, Google says it’ll extend to generating any output from any input.
What makes Omni different from existing video generation tools is its understanding of physics. It actually models forces — gravity, kinetic energy, fluid dynamics — to make generated scenes feel grounded in reality. That’s not a gimmick; realistic physics in generated video has been one of the hardest problems in the space.
Every video created with Omni carries a SynthID digital watermark — imperceptible to the human eye but verifiable through the Gemini app, Chrome, or Search. You can literally ask “Is this AI generated?” and it’ll tell you.
Gemini Omni Flash is available now in the Gemini app and Google Flow for AI Plus, Pro, and Ultra subscribers, and free on YouTube Shorts Remix for users 18+.
Part 2: Google Search Just Got Its Biggest Upgrade in 25 Years
This is not an exaggeration. Google called it that themselves.
AI Mode Hit a Billion Users
AI Mode — Google’s most advanced Search experience — now has over 1 billion monthly users. Queries are more than doubling every quarter since launch. And this is the model they’re now upgrading further, with Gemini 3.5 Flash as the new default.
The Search Box Itself Has Changed
The biggest interface change: the Search box is now multimodal. You can search using text, images, files, videos, or Chrome tabs — and Search will reason across all of them simultaneously.
AI Overviews and AI Mode are merging into a single seamless experience. The flow is: you ask a question → you get a results page with an AI Overview → you naturally continue into AI Mode for follow-up. It’s live right now across desktop and mobile, worldwide.
Search Agents — Always On in the Background
Google is introducing information agents inside Search.
The idea is that instead of you going to Search, Search comes to you. You create an agent for a topic, task, or project. It monitors blogs, news, social posts, and real-time data sources 24/7, and sends you intelligent updates when something relevant happens.
Think of it as a personal research assistant that runs while you sleep. You can spin up multiple agents simultaneously. Rolling out this summer for AI Pro and Ultra subscribers.
Generative UI — Search Builds Custom Interfaces On the Fly
This is genuinely new territory.
With Generative UI, Search doesn’t just return links or AI summaries — it builds custom interface layouts for your specific question. Tables, interactive visuals, graphs, simulations — assembled in real time by Gemini 3.5 Flash.
For ongoing projects like planning a wedding or managing a home move, Search goes further: it can build mini apps, dashboards, and trackers you keep coming back to. These are custom experiences, not pre-built templates.
Generative UI rolls out this summer for everyone, free. Custom app building comes later, starting with subscribers.
Personal Intelligence — Now in 200 Countries
Personal Intelligence in AI Mode is expanding to nearly 200 countries across 98 languages — no subscription needed. You can connect Gmail and Google Photos for a more personalized search experience, with Google Calendar support coming soon.
Part 3: Universal Cart — Shopping Across All of Google
Universal Cart is Google’s bet on becoming the central hub for online shopping.
Here’s the thing: right now, you shop across Search, Gemini, YouTube, Gmail — all separately. Universal Cart unifies those surfaces. Add something to your cart while browsing Search, chatting with Gemini, watching a YouTube video, or reading a Gmail newsletter. One cart, everywhere.
Once you add something, the cart works in the background — tracking price drops, flagging incompatibilities between products, pulling in your loyalty info and payment perks from Google Wallet to help you choose the best checkout option.
Checkout is handled by Universal Commerce Protocol (UCP), which enables one-tap checkout via Google Pay for participating brands, or direct transfer to the retailer’s own site.
Rolling out across Search and Gemini this summer; YouTube and Gmail to follow.
Part 4: Google Antigravity — The Developer Platform That Replaces Everything Else
Google Antigravity is the new name for Google’s agent-first development platform. If you’ve been following Google’s developer tools, this unifies and supersedes what came before.
The core premise: Antigravity lets you build with AI agents, not just AI models. You orchestrate multiple agents that work in parallel — one codes a website while another generates brand assets.
Four things launched this week:
Antigravity 2.0 — a standalone desktop app where you manage and orchestrate multiple agents simultaneously.
Antigravity CLI — a terminal-first version for developers who don’t want a graphical interface. Lightweight and fast.
Antigravity SDK — programmatic access to the same agent infrastructure powering Google’s own products, co-optimized for Gemini models. You can host it on your own infrastructure.
Managed Agents — via the Gemini API, a single API call spins up a remote Linux environment where an agent can browse the web, run code, manage files, and call tools. You extend it with your own instructions written in markdown files.
Multi-day engineering efforts, Google says, are collapsing into hours. The subagent teamwork capability — multiple agents collaborating on a task — is in early research preview now.
Part 5: The Gemini App Gets a Full Redesign
Gemini Spark — Your 24/7 Personal AI Agent
Gemini Spark is the most personal product Google announced.
It’s an autonomous AI agent that works in the background on your phone or laptop — even when the device is off. It runs on Gemini 3.5 and operates under your direction: you turn it on, and it checks with you before taking major actions.
The planned features coming through summer: you’ll be able to text or email Spark directly, create custom sub-agents for specific tasks, and authorize payments within set budgets and merchants.
Currently rolling out to trusted testers. Beta coming to AI Ultra subscribers in the US next week.
Daily Brief — AI That Prepares Your Day Overnight
Daily Brief is an out-of-the-box agent that works while you sleep.
It goes through your inbox, calendar, and tasks to find the most important things for your day, then delivers a personalized digest with suggested next steps. It learns your preferences over time.
Available today for all Google AI subscribers (18+) in the US, in the Gemini app.
Neural Expressive — The App Looks Different Now
Google redesigned the Gemini app from scratch with what they call Neural Expressive — fluid animations, vibrant colors, new typography, haptic feedback. More practically: responses are no longer walls of text. They’re laid out in real time, with interactive images you can zoom into, timelines you can skim, and embedded visuals.
Gemini Live is now inline — it opens immediately, with a newer model that’s faster and better at filtering background noise.
Part 6: Gemini for Science
This one is less consumer-facing but worth knowing about.
Google launched Gemini for Science — a collection of tools for researchers:
Hypothesis Generation — built with Co-Scientist (a multi-agent system), it runs an “idea tournament” where hypotheses are generated, debated, and evaluated. Claims are verified and linked to citations.
Computational Discovery — an agentic research engine built with AlphaEvolve. It generates and scores thousands of code variations in parallel, letting scientists test modeling approaches that would take months to navigate by hand.
Literature Insights — built with NotebookLM, searches scientific literature and structures results into tables with custom, searchable attributes for comparative analysis.
All three are open for access registration at labs.google/science from May 19.
Additionally, Science Skills integrates data from 30+ major life science databases — UniProt, AlphaFold, AlphaGenome, InterPro — directly into Google Antigravity. Tasks that used to take hours of manual bioinformatics work can now happen in minutes.
Part 7: The Other Big Announcements (That Aren’t Small)
Google AI Studio — Build Android Apps in the Browser
Google AI Studio now lets you build native Android apps directly from the build tab. You can preview on an in-browser Android emulator, connect a test device via ADB, and publish to Google Play’s Internal Test Track with one click. The first two apps deploy to Google Cloud at no cost — no credit card required.
Google Workspace Updates
AI Inbox (launched earlier this year) now generates personalized draft replies in Gmail and surfaces relevant Docs, Sheets, and Slides next to your to-dos. AI Pro and Ultra subscribers get Gmail Live this summer — ask conversational questions about your inbox without digging through threads.
Google Pics is a new image creation and editing tool built on Nano Banana (Google’s latest image model). It includes object segmentation, text editing and translation, and Workspace integration. Launching to limited testers now, rolling out globally to AI Pro/Ultra subscribers this summer.
Docs Live — create and edit Google Docs with your voice. The feature pulls context from Gmail, Drive, Chat, and the web. Available this summer for AI Pro/Ultra subscribers.
Google Flow — AI Creative Studio
Google Flow now has an agent. The Flow Agent handles multi-step creative tasks: early brainstorming, creating multiple variations simultaneously, batch editing, and organizing assets. Available to all Flow users globally now.
Google Flow Tools lets you build custom creative tools inside Flow using natural language — no coding required. Create image editors, video resizers, custom shaders, and share them with other Flow users.
Google Flow Music — create music videos conversationally, edit specific sections of songs (change lyrics, genre, instruments), and collaborate with an AI agent that understands music direction.
Ask YouTube
Ask YouTube is a new conversational search experience for YouTube. Instead of keyword search, you ask complex queries — “what’s the best way to teach a kid to ride a bike?” — and it compiles the most relevant long-form videos and Shorts into an interactive, structured response. Rolling out this month to a subset of US users searching in English.
Android XR — Smart Glasses This Fall
Google confirmed two types of intelligent eyewear under Android XR:
Audio glasses — worn on your face, providing spoken AI assistance in your ear. Made in partnership with Gentle Monster, Warby Parker, and Samsung. Compatible with Android and iOS. Arriving this fall.
Display glasses — show you relevant information visually, right in your field of view. Timeline less specific.
SynthID Expanding Everywhere
SynthID — Google’s AI watermarking technology — is three years old now and has been verified 50 million times globally. It’s expanding to Google Search today (via Lens, AI Mode, Circle to Search) and Chrome in the coming weeks.
Google is also adding C2PA Content Credentials verification — this tells you whether content is an unaltered original from a camera or has been modified, and by what tools.
The bigger news: OpenAI, Kakao, and ElevenLabs are now bringing SynthID watermarks to their own AI-generated content, pushing this toward an industry standard.
What This All Adds Up To
A few themes that cut across everything Google announced this week:
Agents, not chatbots: Almost everything Google launched is agent-first. Gemini Spark, Daily Brief, information agents in Search, Flow Agent, Managed Agents in the API — the shift is from AI that answers questions to AI that executes tasks autonomously over time.
AI inside existing surfaces: Google is not trying to make people use new apps. They’re bringing AI into Search, Gmail, YouTube, Google Docs, and Maps. That’s a different strategy than building standalone AI products, and it plays to Google’s existing distribution advantages.
Watermarking is becoming a standard: SynthID started as a Google-only technology. Now it’s in products from OpenAI, Kakao, and ElevenLabs. That’s meaningful — not just for transparency about AI content, but as an emerging industry norm.
The $100 AI Ultra plan targets professionals: The new tier offers 5X usage limits compared to AI Pro, 20TB of storage, and priority access to new features. It’s positioned for developers, technical leads, and advanced creators — not casual users.
TL;DR — What Actually Matters
| Announcement | What It Is | Who It’s For |
|---|---|---|
| Gemini 3.5 Flash | Fastest frontier-quality AI model yet | Developers, power users |
| Gemini Omni | Any input → any output (starts with video) | Creators, media professionals |
| Google Antigravity | Agent-first dev platform | Developers, builders |
| AI Mode upgrade | AI-native Search box with Generative UI | Everyone |
| Universal Cart | Unified shopping across Google | Online shoppers |
| Gemini Spark | Autonomous personal agent, 24/7 | AI Ultra subscribers |
| Daily Brief | Overnight inbox + calendar digest | Google AI subscribers |
| Android XR glasses | Smart glasses (audio + display types) | Early adopters, fall 2026 |
| SynthID expansion | AI content watermarking in Search + Chrome | Content consumers |
| AI Ultra plan | $100/month tier for professionals | Developers, advanced creators |
