Google I/O 2025: Key Highlights, Summary, Latest AI and Other Announcements

Google I/O 2025 took place May 20-21, 2025, at the Shoreline Amphitheatre in California. It was primarily dedicated to artificial intelligence, kicked off by Google’s current CEO, Sundar Pichai. He announced new versions of the Siri AI models Gemini, including 2.0 and Flash.

The models are designed to improve their reasoning capabilities and creativity and move Google closer to the goal of achieving Artificial General Intelligence (AGI).

Buckle up, we’re just getting started 🚗 #GoogleIO pic.twitter.com/qI4NURT1XE

— Google (@Google) May 21, 2025

Among the highlight announcements were the AI Mode in Google Search, providing smarter and more interactive results, and the debut of Project Astra, a real-time AI assistant that understands video and audio inputs. Google has also released Google Beam, a 3D.

In this article, we’ll look at all the major announcements from Google I/O 2025. We’ll cover the latest AI breakthroughs, updates to Google’s core products, and exciting new tools like Flow and Gemini Live.

Major Announcements from Google I/O 2025

At #GoogleIO, we shared how decades of AI research have now become reality.

From a total reimagining of Search to Agent Mode, Veo 3 and more, Gemini season will be the most exciting era of AI yet.

Some highlights 🧵 pic.twitter.com/2n9rbGNj0Q

— Sundar Pichai (@sundarpichai) May 20, 2025

Here are the top major announcements from Google I/O 2025, which include:

1. Big Improvements in Google’s Gemini AI

Google has made huge progress with its Gemini AI models. The Elo scores, which show how good the models are, have gone up by over 300 points since the first Gemini Pro. The latest version, Gemini 2.5 Pro, is now leading in all categories on the LMArena leaderboard.

2. New AI Chips Make Gemini Faster and Cheaper

This improvement is possible because of Google’s advanced AI chips. The new TPU, called Ironwood, is designed for AI tasks and is 10 times more powerful than the previous version. It can do 42.5 exaflops of computing per pod.

3. Google Offers the Best AI at the Best Price

Thanks to these powerful chips, Gemini models are faster and cheaper. Google has stayed ahead by offering strong performance at a lower cost. In fact, they’ve pushed the limits of what’s possible with AI pricing and performance.

4. AI Use Is Growing Fast Around the World

People are using AI more than ever. A year ago, Google was processing 9.7 trillion tokens a month. Now, it’s 480 trillion. That’s 50 times more. Over 7 million developers are using Gemini, five times more than last year. The Gemini app now has over 400 million monthly users, and use of the 2.5 Pro model in the app is up by 45%.

5. Google Beam Makes 3D Video Calls Real

Google introduced Google Beam, a new AI-powered video platform. It turns 2D video into 3D using six cameras and AI. It tracks your head movements with high accuracy and runs at 60 frames per second, making it feel like you're in the same room. Beam will be available to early customers this year in partnership with HP.

6. Live Translations Coming to Google Meet

Google Meet will soon include live translation with voice matching. It will copy the speaker’s voice, tone, and facial expressions. This feature is launching in English and Spanish for Google AI Pro and Ultra users and will expand to more languages soon.

7. Gemini Live Brings AI to Real-World Tasks

Gemini Live now includes features from Project Astra. It can use your phone’s camera and screen share to help with tasks like job interviews or training plans. It’s already available on Android and is rolling out to iOS users.

8. Agent Mode Lets Gemini Do Tasks for You

Google is developing Agent Mode using technology from Project Mariner. It lets Gemini perform tasks on websites, like finding apartments and scheduling tours. This is possible through a “teach and repeat” method. An early version of Agent Mode will be available soon for subscribers.

9. Gemini Gets Personal with Smart Replies

Gemini will soon offer Smart Replies that are more personal. It can search your emails and files to suggest detailed and personalised replies. It even matches your tone and style. This feature is coming later this year for subscribers.

10. Google Search Gets an All-New AI Mode

A new AI Mode in Google Search allows for longer, more complex questions. You can also ask follow-up questions. It’s now available in the U.S. and powered by Gemini 2.5. This new experience is faster and more accurate.

11. Gemini 2.5 Gets a New Thinking Mode

Gemini 2.5 Flash is now even faster and better for reasoning, coding, and long tasks. Gemini 2.5 Pro now includes "Deep Think", a mode that improves how the model reasons and solves problems.

12. New Tools in the Gemini App

The Gemini app now supports file uploads and will soon connect to Drive and Gmail. This helps create personalised research. It's also being linked to Canvas for making infographics, quizzes, and podcasts. Users can now build apps just by chatting with Gemini.

13. Creative AI: New Video and Image Tools

Google launched Veo 3, a video model with audio generation, and Imagen 4, an image generator. Both are available in the Gemini app. Google also introduced Flow, a new tool for filmmakers to create and expand video scenes.

14. Using AI to Make Life Better

Google believes AI can improve lives. From helping scientists to letting people ride in driverless cars like Waymo, AI is creating real change. Google is committed to turning research into tools that benefit everyone.

Latest AI Innovations: Gemini, Project Astra, and More

At Google I/O 2025, Google unveiled major advancements in artificial intelligence, led by its Gemini models. Gemini 2.5 Pro now tops AI performance charts and introduces a new "Deep Think" mode for better reasoning. The faster, more efficient Gemini 2.5 Flash is also gaining popularity, especially among developers.

Project Astra's technology is now part of Gemini Live, which lets users interact with the real world using camera and screen sharing.  On top of that, Agent Mode, developed from Project Mariner, allows Gemini to take actions online, like apartment hunting and scheduling tours.

Other innovations include:

• Google Beam: AI-powered 3D video calls for a more lifelike communication experience.
• Live translations in Google Meet, matching the speaker's voice and tone in real time.
• Smart Replies in Gmail: Personalised responses using your past conversations and writing style.
• A new AI Mode in Search, allowing deeper, more complex queries and follow-ups.
• Creative tools like Veo 3 and Imagen 4, plus Flow, a new AI filmmaking assistant.

Altogether, these innovations show how Google is pushing the boundaries of AI to make it more powerful, personal, and widely available.