Google kicked off its annual I/O developer conference last night, and as expected, the company focused almost exclusively on AI.

Whether that be what new models it was bringing to market, how it wanted to integrate AI into more of its products, as well as building what it is calling a “universal AI”.

With a lot of content to digest in such a short period of time, we;ve rounded the biggest announcements made during the Google I/O 2025 keynote below.

Here’s what caught our eye.

Gemini 2.5 Flash and Pro on the way

Starting with Gemini, and the 2.5 Pro and Flash iterations of generative AI models are expected to be made generally available soon.

No precise date has been mentioned, but Google is bringing new capabilities such as native audio output for a more natural conversational experience, as well as enhanced security safeguards, along with Project Mariner’s computer use capabilities improving.

The company also stated that Gemini 2.5 Pro will get better with Deep Think, which is an experimental, enhanced reasoning mode for highly complex mathematics and coding.

More AI Mode in Search

For users in the US alone, later this week, Google will roll out a dedicated AI Mode tab in Search. The feature will be powered by the Gemini AI chatbot, and looks to be an extension of the AI Overviews that Google brought to the fore late last year.

It’s unclear whether this AI Mode will be better at producing more accurate results than AI Overviews, which got a fair bit wrong in its early months, but Google shows no signs of slowing down in this department, with plans to test out more features within the Mode such as deep search and generating charts based on data found.

Making its AI universal

Google has made no secret about making its AI all-encompassing. To achieve this, the company wants to create a “universal” AI model.

“We’re working to extend our best multimodal foundation model, Gemini 2.5 Pro, to become a “world model” that can make plans and imagine new experiences by understanding and simulating aspects of the world, just as the brain does,” it noted in a blog post.

“Making Gemini a world model is a critical step in developing a new, more general and more useful kind of AI — a universal AI assistant,” it added. There’s no timeframe, but that’s Google’s grander plan for Gemini.

AI Overviews is very much international

Speaking of AI Overviews, despite its initial troubles, Google has pushed ahead with its expansion into more territories. To that end, it is now available in more than 200 countries and supports more than 40 languages.

How successful it has been, however, is unclear, with no data being shared about how frequently search users opt for its information over other results, or how many people have opted to disable it in search.

That said, Google did note that in its biggest markets, “US and India, AI Overviews is driving over 10% increase in usage of Google for the types of queries that show AI Overviews.”

Google AI Ultra is ultra expensive

As the name would suggest, Google AI Ultra is a newly announced subscription plan that’s said to offer, “the highest usage limits and access to our (Google’s) most capable models and premium features.”

It is available in the US only for now, and costs $249.99 per month (~R4 470), and as such, appears to be an enterprise-focused offering.

In terms of the models it has access to, it includes Gemini, Flow, Whisk, NotebookLM, Gemini in Gmail, Docs, and Vids, Gemini in Chrome, Project Mariner, YouTube Premium, and 30 TB of storage.

Shopping in AI Mode

One of the other elements coming to AI Mode is shopping, with Google seeking to keep users on its services instead of heading to the retailer. “The Shopping Graph now has more than 50 billion product listings, from global retailers to local mom and pop shops, each with details like reviews, prices, color options and availability,” it boasted.

It’s also enhancing its virtual try-on, via Search Labs for users in the US only for now. “When you’re shopping for shirts, pants, skirts and dresses on Google, simply tap the ‘try it on’ icon on product listings. From there you can upload a full-length photo of yourself — and within moments, you’ll see how that wedding-season maxi dress or playful shirt for your next vacation looks on you,” it explained.

More Android XR partners

Lastly, while we did not get the fully fledged reveal or demo of Project Moohan we were hoping for, Google did provide an update for its extended reality platform, Android XR.

Being developed in partnership with Samsung and Qualcomm, the Project Moohan XR headset is expected later this year, but Google also wants to find other partners to collaborate with on extended reality.

“We’re partnering with innovative eyewear brands, starting with Gentle Monster and Warby Parker, to create stylish glasses with Android XR. And in the future, we look forward to working with more partners, like Kering Eyewear, to bring even more options to users,” Google shared, with smart glasses being the form factor it’s focusing on right now.

You can also watch a 10-minute recap of the announcements made during the keynote for Google I/O 2025 in the video embedded below.