Google I/O 2026 Opens Tomorrow: Gemini, Agentic Coding, and the Android XR Gap

Google I/O 2026 opens tomorrow at 10:00 AM PT with the main Google Keynote, followed by the Developer Keynote at 1:30 PM PT, both livestreamed for free from Shoreline Amphitheatre. The two-day event runs May 19–20, and the published session list points squarely at Gemini model updates, agentic coding tools, and Android 17, with one conspicuous absence: a dedicated Android XR session.

That omission is the most interesting thing about an otherwise predictable agenda. Android XR doesn't appear explicitly in the sessions list beyond a mention of "Adaptive Everywhere" in the Android 17 entry, and Google has not confirmed whether a dedicated Android XR session will be announced before or during the keynote. For a platform Google has spent the past year promoting alongside Samsung, Warby Parker, and Gentle Monster, that is a strange way to enter your biggest developer week of the year.

Google I/O 2026 keynote time and how to watch

The opening keynote streams at 10:00 AM PT on Tuesday, May 19, with the Developer Keynote at 1:30 PM PT the same day. Google has confirmed the event runs May 19–20 with livestreamed keynotes and sessions, hosted at io.google. No registration is required to watch, and replays land on YouTube shortly after each session airs.

The in-person event is invite-and-lottery only. Google has released the list of sessions for I/O 2026, outlining the main topics planned for the two-day event on May 19 and 20 at Shoreline Amphitheatre in Mountain View, California, with attendance for developers limited to those selected through a lottery or invitation system. For everyone else, the livestream is the entire show.

Four sessions follow the main keynote on day one at 3:30 PM PT. According to Android Authority, the lineup is What's new in Google AI, What's new in Android, What's new in Chrome, and Agent-first workflows from prompt to production. A second block at 4:30 PM PT covers Google AI Studio and Antigravity, Google Play, AI coding workflows on the web, and Firebase.

Gemini 4 at Google I/O is the rumor nobody can pin down

A Gemini 4 reveal is the loudest unconfirmed rumor heading into Tuesday. The session descriptions describe Google's end-to-end AI stack and "latest model capabilities," which is the kind of phrasing that historically precedes a numbered model bump, but Google has not named the model publicly. Google has scheduled a block of highly anticipated sessions, including one titled "What's new in Google AI," and according to the official descriptions, these sessions will explore Google's end-to-end AI stack and the capabilities of their latest models, which could be Gemma 4.

That ambiguity is doing a lot of work. The same language could cover a Gemini 4 frontier model, a Gemma 4 open-weights refresh, or both. Either way, the AI session is scheduled for 3:30 PM PT on May 19 and is expected to cover multimodal capabilities, media generation, and robotics. An AI-focused session is scheduled for 3:30 PM PT on May 19, covering the latest capabilities in multimodal processing, media generation, and robotics.

The consumer-side Gemini story already partially leaked at The Android Show on May 12. According to Engadget's recap of The Android Show, Gemini in Chrome for Android arrived built on Gemini 3.1, with research, summarization, and cross-product reach into Gmail, Calendar, and Keep. An auto-browsing capability inside Chrome is set to roll out at the end of June to devices running Android 12 or later in the US, initially gated behind AI Pro and Ultra subscriptions.

That staged consumer rollout is the clearest signal about what Tuesday's keynote will not do. Google has already taken the obvious Gemini news off the table, which means the I/O keynote needs a new model, a developer hook, or both to justify ninety minutes of stage time. For more on how Google is positioning the broader Gemini Intelligence platform across phones, watches, and cars, see our analysis of Google's Gemini Intelligence bet.

Agentic coding moves from demo to deployment

The most concrete developer story on the schedule is agent-first coding. Day one carries a session titled "Agent-first workflows from prompt to production," sitting next to a second session on building AI applications with Google AI Studio and a new tool called Antigravity. At 3:30 PM on May 19, Google has four sessions lined up: What's new in Google AI, What's new in Android, What's new in Chrome, and Agent-first workflows from prompt to production. Then, at 4:30 PM, another four sessions follow: Build next-gen AI experiences with Google AI Studio and Antigravity, What's new in Google Play, Unlock modern web capabilities in your AI coding workflows, and What's new in Firebase.

Firebase is the most quietly significant rebrand on the agenda. According to Android Gadget Hacks' I/O 2026 preview, Firebase is being repositioned as an "agent-native platform," with a path that runs from AI prototyping through to production deployment on Google Cloud, plus integrations with AI Studio and Antigravity. That is Google explicitly framing Firebase as the runtime for autonomous agents rather than a mobile backend.

There is a guardrail story riding alongside the agent push. Engadget reported that Google confirmed agentic Gemini Intelligence features require user confirmation before executing sensitive tasks like purchases or social media posts, a direct response to concerns about unsupervised AI agents. The phrasing matters: Google is not promising agents will never act unsupervised, only that current sensitive categories are gated. Developers building on Antigravity should expect that list of "sensitive" actions to shift, possibly multiple times before stable APIs ship.

The coding workflow angle is where Google is most exposed to comparison. Anthropic, OpenAI, and Microsoft have all shipped agentic coding products in the past year, and Google is arriving with a tool most developers have not heard of. For context on why agent platforms keep colliding with security issues, our reporting on vibe-coded apps is useful background.

The Android XR smart glasses 2026 problem

Google's biggest hardware bet of the past twelve months does not have a session slot. Surprisingly, there's no word on a dedicated Android XR session. Considering Google is already working on XR glasses with Android XR, it's possible that the brand may talk about it in the Android section.

That is unusual. One year ago at I/O 2025, Google announced smart glasses partnerships with Warby Parker and Gentle Monster, positioning Android XR as the platform answer to Meta Ray-Bans and Apple's rumored glasses program. Six weeks before I/O 2026, Google shipped a substantial Android XR update with auto-spatialization for 2D apps, persistent wall-anchored apps, and Android Enterprise support across MDM partners including Microsoft Intune and Samsung Knox Manage. According to VR.org's pre-I/O preview, Google has confirmed that Android XR glasses will be available for hands-on demos at I/O, the first time developers and press will physically interact with the hardware outside controlled briefings.

So the hardware is real, the platform is shipping monthly updates, and the demos are scheduled. But the public agenda buries XR inside an Android 17 talk on "Adaptive Everywhere." The following day, there is a separate session focusing on Android 17's "Adaptive Everywhere" approach. This initiative aims to enable users to seamlessly transition between phones, cars, living rooms, and immersive environments. Google has indicated that this strategy extends across Android, ChromeOS, and XR, all under a unified platform.

The most likely explanation: Google is saving the XR announcement for the main keynote stage rather than a 30-minute developer breakout. The two-day event will be livestreamed and is expected to cover updates across Gemini, Android, Chrome, and more, with keynotes, demos, and what Google calls "Dialogues" sessions featuring conversations about where AI is headed. If Google has new glasses to show, the May 19 keynote is the moment, and the breakouts will fill in afterward. If it doesn't, Tuesday will be the first I/O in three years where XR is more of a footnote than a headline. Our best AR/VR headset buying guide covers where Samsung's Galaxy XR currently sits against Apple and Meta.

What got pushed to The Android Show on purpose

Google ran The Android Show: I/O Edition on May 12, one week before the main keynote, for the second year running. The pre-event has become a deliberate offload valve. The biggest reveal was Googlebooks, a new laptop line extending the Chromebook family on the 15th anniversary of the original Chromebook. Engadget's writeup covers the full slate, including Magic Pointer, a gesture-based cursor built with Google DeepMind that triggers contextual Gemini suggestions by wiggling at on-screen content.

That scheduling choice tells you what Tuesday is actually for. By moving consumer Android, Chromebook hardware, and the Gemini in Chrome announcement to the prior week, Google has carved out keynote real estate for platform AI and developer tooling. A 45-minute Android 17 session remains on the I/O schedule, but Beebom notes the slot is positioned to add depth rather than introduce new features.

The one historical pattern worth tracking: Google typically does not publish every session before the keynote. Google I/O 2026 livestreams May 19 and 20, with keynotes and sessions available online, confirmed on the Google I/O site. Based on Google's established I/O format, sessions are typically broadcast through the I/O website and YouTube simultaneously, with replays available after the fact. Specific keynote start times, regional scheduling, and direct watch URLs are not confirmed in current sourcing. The XR session may still appear after Sundar Pichai walks off stage, in which case the omission was a marketing choice, not a strategic retreat. More News articles tracking the I/O announcements as they land will publish across May 19 and 20.

What to watch at 10:00 AM PT on May 19

Three specific things will tell you within the first thirty minutes of the keynote whether I/O 2026 is the AI platform year Google needs it to be:

• Whether the opening segment names a new flagship Gemini model with a version number, or restates Gemini 3.1 as the current frontier
• Whether Android XR glasses appear on the main stage with a developer availability date attached, not just a demo booth
• Whether Antigravity is positioned as a standalone product or a feature inside AI Studio, which determines how seriously Google is treating the agentic coding category

The livestream is free at io.google/2026. The dates were announced on February 17, 2026, roughly three months before the event, per Wokeey's event guide, and the in-person slots filled within hours of registration opening. Everyone else gets the same view as the developers in Mountain View.