Google spent its biggest developer event of the year arguing that artificial intelligence should act, not just answer. At Google I/O 2026, held May 19 to 20 in Mountain View, the company unveiled a new model family, a personal agent that runs around the clock and an intelligent shopping cart, alongside what it called the deepest change to Search in the product’s history. CEO Sundar Pichai described the moment on stage as the start of the agentic Gemini era, and the word agent was attached to nearly every product shown.
The shift matters now because of how fast Google’s usage is climbing. Pichai said the Gemini app has crossed more than 900 million monthly users, up from 400 million a year earlier, with daily requests growing roughly sevenfold and the model APIs now processing around 19 billion tokens per minute. He said Google expects capital spending of about $180 billion to $190 billion this year, up from $31 billion in 2022, much of it going toward its eighth-generation TPU 8t and TPU 8i chips for training and inference.
New models lead the lineup
The headline release was Gemini 3.5 Flash, the first model in the Gemini 3.5 family. It became generally available on the opening day through the Gemini app, AI Mode in Search, the Gemini API, the Gemini Enterprise Agent Platform and the Antigravity development platform. Google says the model outperforms the earlier Gemini 3.1 Pro on demanding coding and agentic tests, scoring 76.2% on Terminal-Bench 2.1 and 83.6% on MCP Atlas, while running about four times faster than other frontier models. A larger Gemini 3.5 Pro is already in internal use and is expected to roll out next month.
Google also introduced Gemini Omni, a separate model designed to generate output from any input, starting with video. Google says the model pairs Gemini’s reasoning with a sharper grasp of physics, including gravity, kinetic energy and fluid dynamics, so generated scenes look more realistic. Videos created with Omni carry Google’s imperceptible SynthID watermark. Gemini Omni Flash is rolling out now to all Google AI Plus, Pro and Ultra subscribers worldwide through the Gemini app and Google Flow, and it is also free in YouTube Shorts Remix for users 18 and older.
Search gets its biggest change in 25 years
Google called the new Search box the biggest upgrade to that surface in over 25 years. It lets people search using text, images, files, videos and Chrome tabs, with Search reasoning across all of them at once. The company is also folding AI Overviews and AI Mode into one seamless experience. AI Overviews now reaches more than 2.5 billion monthly users, and AI Mode, powered globally by Gemini 3.5 Flash as its default model, has passed 1 billion.
The bigger story is the arrival of Search agents. Google is starting with information agents, which work in the background 24/7 to monitor a topic across the web and send back synthesized updates. They roll out this summer, first to Google AI Pro and Ultra subscribers. Search can also build custom layouts on the fly, a feature Google calls generative UI, which arrives for everyone at no cost this summer, with custom dashboards and trackers following in the months after. Separately, Google is expanding Personal Intelligence in AI Mode to nearly 200 countries and territories across 98 languages, letting people connect apps such as Gmail and Google Photos, with Calendar support coming.
The Gemini app turns into an agent
Inside the Gemini app, Google introduced Gemini Spark, a 24/7 personal agent that takes action on a user’s behalf under their direction. Spark runs on dedicated cloud machines, so it keeps working even when a phone or laptop is switched off, and it is designed to check in before taking major actions. Google is rolling it out cautiously, starting with trusted testers and a beta planned for Google AI Ultra subscribers in the US next week.
A companion feature called Daily Brief works overnight, analyzing a user’s inbox, calendar and tasks to assemble a prioritized digest, rolling out from launch day to Google AI subscribers in the US who have connected their apps. Google also rebuilt the app’s interface with a new design language called Neural Expressive, replacing walls of text with interactive images, timelines and embedded visuals as the response streams in.
A unified platform for builders
For developers, Google expanded Antigravity into version 2.0, a desktop application that can orchestrate several agents in parallel, alongside a command-line interface and a software development kit. Google also said it is transitioning the older Gemini CLI into the new Antigravity CLI, and on June 18 the legacy tool will stop serving requests for Google AI Pro and Ultra and free Gemini Code Assist accounts, though enterprise access is unaffected. The company launched Managed Agents in the Gemini API, where a single call provisions a remote Linux environment in which an agent can plan, run code and browse the web. Google AI Studio gained native Android app building, and Google announced a $100 AI Ultra plan for developers, with usage limits five times higher than the Pro tier, plus a Build with Gemini XPRIZE hackathon carrying a $2 million prize pool.
Shopping, work and creative tools
Google introduced Universal Cart, a shopping cart that looks for deals, tracks price history and flags product incompatibilities once an item is added. It rolls out across Search and the Gemini app this summer, with YouTube and Gmail to follow. Across Google Workspace, the company expanded AI Inbox in Gmail and added voice-driven editing to Docs and Keep. It also launched Google Pics, an image creation and editing tool, starting with trusted testers before reaching Google AI Pro and Ultra subscribers globally this summer. Creative platform Google Flow gained an agent that can plan multi-step projects.
Google also introduced Gemini for Science, a set of research tools for generating hypotheses and structuring scientific literature, and showed that DeepMind’s Project Genie can now build interactive virtual worlds from real Street View imagery for subscribers on the $200 monthly plan. On hardware, the next milestone for Android XR is intelligent eyewear, with the first audio glasses, made with Gentle Monster, Warby Parker and Samsung, arriving this fall. Google also detailed Ask YouTube, a conversational discovery experience now in testing, and expanded SynthID and C2PA content verification to Search and Chrome.
Google I/O 2026 made clear that Google wants Gemini to become less of a chatbot and more of an operating layer across Search, shopping, work, creativity and software development. The strongest announcements were Gemini 3.5 Flash, Gemini Omni, Search agents, Gemini Spark, Universal Cart and Antigravity’s expansion into an agent platform. The real test is availability and trust. Many of the most ambitious features are limited to subscribers, trusted testers, US betas or summer rollouts, so what shipped on stage was a model and a direction more than a finished product. Google still has to prove that agents acting across inboxes, browsers, shopping carts and documents can be useful without becoming intrusive.
Read the full article here
