Major announcements made at Google I/O 2025

A day after Microsoft kicked off its annual Build conference in Seattle, tech rival Google unveiled its flagship Google I/O event 1,300 kilometers south, at the Shoreline Amphitheatre in Mountain View, California.

Google showcased a range of announcements across the Google portfolio related to Artificial Intelligence (AI) chatbot Gemini, Google Search, Android, and Chrome, among others. Here is the list of the key highlights.

Gemini 2.5 with Deep Think

Google announced updates to its Gemini 2.5 Pro model with a new enhanced reasoning mode called Deep Think. Built on Google’s latest innovations in cognitive modeling and parallel reasoning, it takes complex problem-solving to an advanced level, the company said. 

The company’s assessment of this version has shown that 2.5 Pro Deep Think scores impressively on math benchmarks, competition-level coding, and multimodal reasoning. 

Project Mariner

The early research prototype of Project Mariner, which is focused on human-agent interaction, was launched in December. Google has now added new capabilities to the project, including multitasking and a “teach and repeat” method that learns from user actions. Developers will be able to leverage this solution through the Gemini API, with early access already in the hands of partners like Automation Anywhere and UiPath. Google said a broader rollout is coming this summer. 

It supports open standards like the Agent2Agent Protocol and now offer compatibility with Anthropic’s Model Context Protocol (MCP) through the Gemini API and SDK.

Agentic features are also available on Chrome, Search, and the Gemini app. A new Agent Mode, launching soon for subscribers, can help with tasks like apartment hunting, using MCP to search listings, tweaking filters, and scheduling tours.

AI mode in Search

Building on its AI Overviews feature in Search, Google said that it has now introduced AI Mode in Search through Labs. AI Mode is Google’s most advanced search yet, combining reasoning, multimodal understanding, and the ability to dive deeper with follow-up questions and curated web links. In the coming weeks, a dedicated AI Mode tab appear in Search and right in the Google app’s search bar.

Google began testing the AI Mode feature earlier this year, and now it is being rolled out first in the US; users can use it without any sign-up.

Additionally, Google is bringing Gemini to Chrome, introducing a new AI-powered browsing assistant designed to help users quickly grasp a page’s context and complete tasks more efficiently.

Other updates

Google also unveiled a range of other capabilities and solutions, including Beam 3D teleconferencing, Wear OS 6, new AI features in Android Studio, and Project Astra Glasses.