At Google I/O, Sundar Pichai, Demis Hassabis and others took to the stage to outline a long lineup of AI-powered products and services including Gemini 2.5,  AI Mode in Search, which is already being rolled out in the United States and Projects Astra and Mariner.

Gemini 2.5, the latest version of the Google AI assistant that was originally called Bard, is claimed to represent a significant leap forward in AI capabilities, with a focus on deeper reasoning, enhanced multi-modal understanding, superior coding performance, and a more natural and integrated user experience. It has two versions, Pro which is Google's most advanced reasoning model targetted at solving complex problems; and Flash which is optimized for speed and efficiency. Both have already been available for some weeks with Gemini 2.5 Pro Experimental initially released in March 2025 and Gemini 2.5 Flash Experimental following in April. General availability for both models is scheduled for June.

A newer version of Gemini 2.5 Flash was released at the Build event and news came of an experimental "Deep Think" Mode, an enhanced reasoning mode in Gemini 2.5 Pro that allows it to consider multiple hypotheses before responding. This employs new research techniques to provide more robust and accurate answers, particularly for highly complex math and coding problems. This feature is currently rolling out to trusted testers.

Build also gave us a preview of Project Astra with Demis Hassabis introducing it with

This is our ultimate vision for Gemini app, to transform it into a universal AI assistant. In Project Astra we've upgraded voice output to be more natural with native audio. We've improved memory and added computer control. 

If what we see in the video really does work in practice this really will be transformative for day-to-day chores that require gathering information and sourcing parts and materials. 

Gemini 2.5 is the fundamental element of what Google refers to as AI Mode in Google Search. Over time,Google intends to graduate many of AI Mode's cutting-edge features and capabilities directly into the core search experience resulting in:

a total reimaginíng of search.

In his introductory remarks Sundar Pichai told the audience at Build:

Google Search is bringing generative AI to more people than any other product in the world. In our biggest markets like the US and India, AI overviews are driving over 10% growth in the types of queries that show them. What's particularly exciting is that this growth increases over time 

For AI Mode, Gemini 2.5 is going to be augmented by Project Mariner. Developed by Deep Mind this is an AI agent that can browse the internet and perform tasks on your behalf. By integrating Project Mariner into AI Mode, Google Search will be able to not just provide information, but also take action based on your requests. According to the presentations at build this means you could ask AI Mode to perform tasks like finding and purchasing event tickets, booking a restaurant table and comparing prices across multiple shopping sites. The help could extend to filling out forms and researching information and then organizing it into reports or visuals.

AI Mode is already being made available in the United States and from this summer its capabilities will extend to complex analysis and data visualization for sports and financial questions.

More Information

I/O 25

Related Articles

Be Ready For Google I/O 2025

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: [email protected]