Google I/O 2026 as it happened — Gemini Spark, Samsung XR glasses, and everything else announced at Google's giant software showcase

Welcome to our Google I/O 2026 live blog! Stick with us as we predict what to expect from today’s big software showcase, before reporting on the event as it happens.

As a reminder, Google's livestream kicks off at 10am PT / 1pm ET / 6pm BST (or 3am AET tomorrow if you're reading from Australia), and you can tune in via the video link above.

I/O keynotes tend to run for between one and two hours, so expect things to be wrapped up by 12pm PT / 3pm ET / 8pm BST / 5am AET.

Is it Android XR time?

Google I/O 2026 predictions: smart glasses central with XREAL, Gentle Monster & More - YouTube

Watch On

Google has been teasing its Android XR smart glasses for what seems like forever at this point, but might we finally see them unveiled at I/O 2026? We make our predictions in the video above.

Google-what?

The Googlebook was a real oddity of Google's The Android Show broadcast. Is it hardware? Is it software? Google's first Gemini-centric platform is, seemingly, a bit of both, but we're not entirely sure how it works just yet. Here's hoping we get a little more color on this new "intelligence system" at I/O 2026.

Google I/O is typically reserved for software announcements, so we're not expecting much in the way of hardware news from today's event. But if you are wondering about the company's latest physical products, Google recently launched a super slim screenless fitness tracker, the Fitbit Air.

As for where Google I/O is taking place, it'll kick off — like always— at the Shoreline Amphitheatre in Mountain View, California. That's basically Google's version of the Steve Jobs Theater, which, by the way, is just a 16-minute drive away. Keep an eye out for Apple CEO John Ternus peeking over the fence.

Just

30 mins to go!

*Siren noise* Google I/O 2026 kicks off in half an hour. As a reminder, you can tune into the event live via Google's YouTube channel or follow along with us here (if you do hop over to YouTube, you'll be greeted by some rather upbeat, jellyfish-inspired 'jellectronica', which is certainly a choice on Google's part).

It looks like Google CEO Sundar Pichai and co. are ready to rumble...

OK, so while we wait for the main I/O stream to kick off, it looks like we're watching... Pokimane play Infinite Scaler? That's not a sentence I thought I'd be writing this morning.

And we're off! Cue the emotive montage...

Sundar Pichai takes to the stage...

Sundar opens with a run-through of all the ways Gemini has helped to advance fields such as science, education, health, and more in the last year.

Did Sundar just say 'tokenmaxxing'? Yes, yes I think he did.

Google now has 13 products with over a billion users each. Five of those products have over 3 billion users, and AI overviews now have 2.5 billion monthly users. Talk about big numbers.

Here's our first big feature announcement of the day: Ask YouTube.

With Ask YouTube, rather than searching for a specific video the old-fashioned way, you can ask complex and lengthy questions, and Gemini will serve up specific videos that it thinks best answer your query. Helpfully, you'll be sent directly to the relevant part of the videos in question, too, rather than having to skim through them.

Did Google just turn YouTube into a chatbot?

"How my brother inspired me to be a software engineer" feels like a distinctly Google phrase.

That quote came as part of the Docs Live announcement. Rolling out for Google AI Pro and Ultra subscribers, Docs Live will essentially let you speak documents into existence. I'm not worried for my job, you are...

We're now getting a look at the incredible speed of Google's new TPU 8 chips.

Yes, you can now build 8-bit games in about 8 seconds.

Damis Hassabis is now on stage to talk about advancements in Google's AI models. "Artificial general intelligence is just a few years away," he says. As a reminder, that's a theoretical form of AI that can successfully learn, reason, and perform any intellectual task a human can.

Gemini Omni is Google's new AI model. It can simulate complex concepts like kinetic energy and gravity, and translate complex scientific ideas into digestible videos.

The first model in the Omni family is Omni Flash. It's available today in the Gemini app, Google Flow, and on YouTube Shorts.

Google is rolling out SynthID and C2PA verification to Search and Chrome, so users can more easily identify AI-generated imagery.

Next up: Gemini 3.5 Flash. "When compared to 3.1 Pro, Flash is better across the board. It's made huge progress in coding," Sundar says. "It's 4x faster than other frontier models."

Google is supposedly processing more than three trillion tokens a day internally using Gemini 3.5 Flash. Yes, three trillion.

Antigravity CLI, Antigravity SDK, and Native Voice Support are available globally starting today.

Google also announces its Antigravity 2.0 desktop application. It's "unabashedly agent-first," and is supported by Gemini 3.5 Flash.

I'll be honest, guys, I'm not 100% sure what's going on here — Google just name-dropped about seven acronyms in 30 seconds — but I think we're seeing whether Antigravity 2.0 can run Doom.

OK, good news: Antigravity 2.0 is available globally, for everyone, starting today.

This feels big: Google just announced Gemini Spark. This is a personal AI agent that helps you navigate your digital life, taking action on your behalf. It's works 24/7, even when your laptop is closed. Spark is powered by Gemini Flash 3.5 and the Antigravity harness.

We're now getting a Gemini Spark demo (in a redesigned Gemini interface, by the way). The host has tasked Spark with drawing up a complex block party plan involving schedules, planning permissions, and calendar integrations.

Hang on, is that an iPhone?!

Google launches a new AI Ultra plan starting at $100 per month.

Here's Liz Reid now to talk about AI overviews, which have doubled every quarter since last summer. Yikes.

Google Search will now run on Gemini 3.5, and Google is launching an entirely new Search box. Search will "help you formulate your question," and field follow-up queries in a dedicated, chatbot-style box underneath your initial results.

Google's new Search tool can also act like an AI agent. You can ask super complex questions and have it update you outside of Search as the answer to that question changes. Reid gives the example of, "Keep me updated when my favorite athletes drop new sneakers."

"Whether you want to find it, buy it, or book it, Search will help you get it done."

Well, this is pretty mad. Search can also now build custom graphics to help you visualize complex problems. It's called "Generative UI for Search", and it'll be free for everyone starting from summer this year.

If I'm understanding this correctly, you can essentially build mini-apps with Generative UI in Search. The host is giving the example, 'Build a weekend planner to automate my family's weekend plans,' and Search is seemingly building a whole darn app in real-time. This is impressive.

Here's a bit more color on that Ask YouTube feature Google announced earlier.

One query example given by Google was, “How to teach my 3-year-old how to ride a pedal bike, they already know how to ride a balance bike?”

As you can see in the video below, Gemini delivered written answers like a typical AI chatbot, but these were accompanied by relevant YouTube videos, so you can both read and watch to get the answers you’re looking for.

You can read more about Ask YouTube elsewhere on TechRadar.

Onto shopping (yay!). Google has partnered with the likes of Amazon, Shopify, and Walmart on the Universal Commerce Protocol (UCP), which is an open standard designed by Google to unify digital commerce.

Previously, AI assistants had to work with hard-coded integrations for each merchant's unique API, but the UCP levels the playing field to make AI-powered shopping easier.

If that last part was too boring for you, here's something more interesting: the Universal Cart.

This is Google's “new agentic hub for shopping across Google” and “a truly intelligent shopping cart." It comes with AI features powered by the company's Gemini AI model that could help you score a discount or avoid mistakes with the items you’re purchasing.

And here's that big Gemini app redesign. It's called Neural Expressive, and features new colors, animations, and a completely repositioned Gemini Live icon.

Remember that block party prompt Google mentioned earlier? Gemini Spark has just served up the result, and it looks... much better than something I'd be able to produce myself in 30 minutes.

Is Gemini the new GOAT?

Here's a roundup of the many, many Gemini upgrades we just heard in that section.

There was the Gemini on macOS update, the Neural Expressive redesign, the new Omni model, the new 3.5 Flash model, the Daily Brief upgrade, and of course, Gemini Spark.

All of these features bring Gemini one step closer to being "the ultimate personal assistant," as Google describes it.

We're moving away from Gemini now, and look away, graphic designers: Google Pics and Stitch are two new creative tools that make Canva look as advanced as black-and-white television.

Google Flow is also being upgraded with Gemini Omni, new agentic tools, and music-making capabilities. We're seeing an example of a user-generated piano riff being turned into an R&B track. It's not my cup of tea, but you've got to respect the technology.

It's finally Android XR time: Google's first audio glasses will arrive this fall.

These glasses will deliver information directly to your ears, rather than displaying it on your screen, and offer various assistive features, including navigation, summarized notification readouts, real-time audio translation, and the ability to translate text on signs, among other AI tools. You can also use them to capture first-person photos and videos.

Google and Samsung are showing off two glasses designs on stage at I/O, but we'll supposedly get more when the full range launches later this year.

We're now getting a live demo of the glasses in action. They're being used to order a coffee, summarize messages, and add events to the wearer's calendar.

They'll also come with Nano Banana capabilities — so essentially, you can take pictures with the glasses, and tell the on-board AI how you want those pictures to be edited. It's all pretty neat.

Here's some refreshingly positive AI-related news: Gemini for Science will bring together powerful AI tools to assist with research and help scientists model complex concepts.

"This technology will be a force multiplier for human ingenuity and usher in a new age of progress."

And that's a wrap! If you managed to keep up with everything announced in that nearly two-hour showcase, kudos to you — if not, I'll be checking back over my notes to bring you a roundup of the key news imminently.