Frequently asked questions
What is Gemini 3.5 Flash and how does it differ from previous versions ?
What is Antigravity 2.0 for ?
What does Gemini Omni offer compared to other AI video generators ?
What is Gemini Spark and how can you access it ?
Google I/O 2026 will likely be remembered as the moment Google stopped treating artificial intelligence as just another feature and made it the central engine of its entire ecosystem.
The message was clear from the first minutes of the keynote: Gemini is no longer an assistant; it is now the execution layer for Search, Workspace, Android, YouTube, and most of the company's consumer products.
The first announced model is called Gemini 3.5 Flash, and it ushers in a new family of models that Google presents as a breakthrough compared to the previous generation. On paper, it surpasses Gemini 3.1 Pro in coding and agentic execution benchmarks, notably Terminal-Bench 2.1 where it achieves 76.2% and MCP Atlas at 83.6%, all while running four times faster than competing models of the same rank in terms of tokens generated per second.
What truly changes for developers is the introduction of Managed Agents via the Gemini API. A single call is now enough to provision a complete agent with an isolated Linux environment, reasoning capabilities, code execution, and persistent state across sessions.
This is a significant paradigm shift in how AI-powered applications are built. Gemini 3.5 Flash is available immediately in the Gemini app, Google Search, Antigravity 2.0, and via the Gemini API.
The second major model announcement is Gemini Omni, which operates on a different register. Where 3.5 Flash focuses on speed and agentic execution, Omni aims to take multimodality to the next level: simulation.
It accepts text, images, audio, and video as input, and produces video content rooted in a real understanding of the physical world, with a consistency of scene, physics, and movement that typical video generators still struggle to reproduce.
Specifically, rather than generating a clip from a single prompt, Omni allows for iterative work, modifying a video sequence through natural language exchanges, much like working with an editor.
The first version, Gemini Omni Flash, is already available in the Gemini app, Google Flow, and YouTube Shorts.
Antigravity 2.0 is arguably the most foundational announcement for developers.
Google positions it as the go-to platform for designing, testing, and deploying autonomous AI agents, from initial concept to production-ready application.
The new version comes with a centralized desktop application that allows for orchestrating multiple agents simultaneously, as well as an Antigravity CLI command-line tool tightly integrated into this work environment.
Google AI Studio, which directly interfaces with Antigravity, also includes native Kotlin support for Android, one-click deployment to Cloud Run, and a direct connection to Google Workspace APIs.
For Google Cloud enterprise customers, Antigravity can also be directly connected to their cloud projects, with a guarantee that agent data and activity remain within the secure boundaries of their infrastructure.
To rally the developer community around this platform, Google is also launching the Build with Gemini XPRIZE hackathon, with a prize pool of two million dollars.
On the consumer side, the most strategic announcement is Gemini Spark.
It's a 24/7 personal agent capable of operating in the background even when the user has closed their computer or locked their phone. It relies on dedicated virtual machines in Google Cloud and integrates with Gmail, Calendar, and connected tasks to, for example, prepare a personalized morning summary of the day's priorities, suggest actions, and anticipate needs.
Spark will initially launch in beta for Google AI Ultra subscribers in the United States. Google also plans to integrate it directly into Chrome eventually.
Beyond the product announcements, Google also released usage data that illustrates the scale of the ongoing shift.
AI mode in Google Search has surpassed one billion monthly active users, with query volume doubling every quarter, and the Gemini app has exceeded 900 million monthly users.
On the infrastructure front, the new eighth-generation TPUs enable training models at the scale of one million chips worldwide, and Google announced infrastructure investments of $180 to $190 billion for 2026.
What Google I/O 2026 ultimately outlines is a company that no longer needs to convince anyone that AI is important; it is now busy making AI indispensable in every place where its billions of users already spend their time.
