Google I/O 2026

8th Generation TPUs

Google introduced TPU 8t for large-scale training and TPU 8i for low-latency inference. Both deliver major performance gains and are positioned for the next wave of AI infrastructure.

Gemini 3.5 Flash and Gemini 3.5 Pro

Gemini 3.5 Flash is built for coding, autonomous agents, and long-running tasks. Gemini 3.5 Pro is scheduled to launch in the following month for general availability.

Gemini Omni

Gemini Omni is a multimodal model family for text, image, audio, and video. Its early variants, Omni Pro Vision, Omni Pro Audio, and Omni Flash, appear in Search, Gemini, and developer products.

Google Search Agents and Generative UI

Google Search is moving toward agentic search. New Search Agents can monitor changes, surface relevant insights, and reduce manual research. Generative UI allows Search to render dynamic mini-apps directly in the results experience.

Gemini Spark

Gemini Spark is a 24/7 personal AI agent that runs on Google Cloud virtual machines. It supports multi-step tasks, scheduled work, and memory across sessions.

Google Antigravity 2.0

Antigravity 2.0 is Google's agent-first development platform, now available as a standalone product with dynamic subagents, scheduled tasks, and an SDK. It is becoming Google's primary surface for agentic development.

Workspace, YouTube, and Maps AI updates

Google added more conversational AI across products. Ask YouTube can summarize videos, Docs Live can run real-time meetings with AI summaries, Gmail can compose voice-driven messages, and Maps is moving toward navigation that handles natural-language route requests.

Universal Cart and agentic commerce

Universal Cart lets users collect products from Search, Gemini, YouTube, or Gmail and complete purchases in one place. UCP and AP2 are Google's new agentic commerce protocols that allow AI agents to shop, compare, and check out across merchants.

Google AI Studio for Android apps

Google AI Studio now supports AI-assisted Android app creation using Kotlin and integrates directly with Antigravity for a continuous prompt-to-deployment workflow.

Google Pics, Flow, Flow Music, and Stitch

Google expanded creative tools across design, video, music, and UI prototyping. Google Pics is a new image generation app, Flow and Flow Music add cinematic video and original-music creation, and Stitch is a UI prototyping tool that integrates with Antigravity.

Neural Expressive Gemini redesign

Google is redesigning Gemini with Neural Expressive, replacing long chatbot-style responses with rich layouts, custom designs, and visual outputs that feel closer to real digital experiences than traditional chat sessions.

Android XR and intelligent eyewear

Android XR is becoming a physical-space layer for Gemini. Display glasses focus on captions, translations, and turn-by-turn navigation. Smart glasses, including Google Halo glasses, focus on contextual AI assistance, music, and accessibility features.

Safety, science, and trust signals

Google said SynthID has watermarked 100 billion images, with OpenAI and ElevenLabs listed as new partners. Search and Chrome are adding content credentials verification, while CodeMender and Code Security Agent are being made available through APIs to find and fix vulnerabilities in codebases. Google also highlighted WeatherNext and AlphaFold-powered drug discovery work moving into pre-clinical stages.

Rollout timeline

Available now: Gemini Omni Flash, Gemini 3.5 Flash, Antigravity 2.0, Neural Expressive design, Stitch updates, and Flow / Flow Music updates. Coming next: Gemini Spark trusted testing this week, U.S. Google AI Ultra beta access next week, Gemini 3.5 Pro next month, and Search Agents, generative UI, Universal Cart, Ask YouTube, Docs Live, Google Pics, and Spark for Chrome this summer.

• Google Antigravity 2.0 is now positioned as a standalone development layer for running multiple agents across local projects, cloud environments, work trees, and enterprise workflows.
• The platform now supports dynamic subagents, scheduled tasks, a unified CLI, and an SDK for teams that need more control over agent orchestration.

• Managed Agents in the Gemini API let developers create an agentic environment with a single API call.
• Google manages the remote Linux sandbox, secure execution, scaling, and infrastructure layer. Developers provide custom instructions, connected data, and Markdown-defined tools.
• This matters because it reduces the backend work usually needed to deploy autonomous agents.

• Gemma 4 was presented as Google's open model family for agentic tasks, advanced reasoning, and multi-step workflows.
• Google also positioned Gemma 4 for edge deployment, including offline mobile use, robotics, and satellite environments.

• Google AI Studio now supports faster movement from prototype to production.
• New capabilities include Cloud Run deployment, Firebase and Firestore integrations, native Android app generation with Kotlin, and export to Antigravity.
• This makes AI Studio more than a prompting tool. It is becoming a faster build-and-deploy environment for developers.

• Android development is also moving deeper into agent-assisted workflows.
• The Android CLI now integrates with Antigravity, giving agents better access to Android SDK tools, project context, and migration workflows.
• Google also introduced Android Skills and an Android Knowledge Base to help agents follow current Android development practices.

• Google also introduced updates for the agentic web.
• WebMCP was presented as a proposed browser standard that lets developers expose JavaScript functions as tools for browser-based agents.
• Chrome DevTools for Agents gives agents a way to validate, debug, and improve runtime code through tools like Lighthouse audits.