Welcome to the Google IO recap. This will be a recap of the technology advancements and products Google has announced for their AI setup, Chrome, and Google Search. Let’s take a dive and check out what Google has announced.
Gemini
Speaking about the Gemini 2.5 Pro, it dominates the LM-arena leaderboard in all the categories, thanks to the rapid progress in coding. The updated 2.5 Pro is ranked number one in the web development arena, surpassing the previous version by 140 points.
The seventh-generation TPU Ironwood delivers 10x the performance over the previous generation and packs an impressive 42.5 exaflops per pod, and it’s coming to Google Cloud customers this year. The infrastructure down to the TPU is what helps Google deliver the fastest AI models like 2.5 Pro and Flash on top of the LM Arena leaderboard. Gemini holds the top three spots for the highest output tokens generated per second, while the prices of the models have been coming down significantly, and Google has been able to deliver models at the most effective price point.
Today, Gemini Live has Project Astra’s camera and screen-sharing capabilities, so you can talk about anything you see. It’s rolling out to everyone on iOS and Android starting May 20th.
There’s also an agent mode that finds listings from Zillow to match the criteria for the search and uses Project Mariner when needed to adjust very specific filters. Gemini uses MCP to access the listings, and it will keep searching as long as you want, freeing you up to do the stuff you want to do. It’s good for Zillow, which brings new customers and a better conversion rate. The experimental version of agent mode will be coming soon to subscribers.
The Gemini 2.5 flash will generally be available in early June, and pro soon after, and you can try out the preview on Vertex AI, Google AI Studio, Gemini app.
On the Gemini 2.5 flash, they are also previewing text-to-speech, and these now have first-of-their-kind multi-speaker support for two voices built on native audio output and have support for over 24 languages. They can easily go between languages, and you can use this native audio output in the Gemini API from 21st May.
2.5 Flash is even more efficient, reducing the number of tokens the model can use for the same performance, resulting in 22% efficiency gains. 2.5 Flash has thinking budgets to give you control over cost and latency versus quality, and 2.5 Pro will get thinking budgets pretty soon.
Google’s asynchronous coding agent, Jules, which, if u submit a task, takes care of the rest. It integrates with GitHub and works on its own. It can tackle complex tasks in large code bases that used to take a lot of time, like updating the version of Node.js. It can plan the steps, modify the steps, and more in minutes. Jules is in public beta, and anyone can sign up via jules.google
Starting this week, Gemini 2.5 is coming to search and the same model to power the AI overview
They are bringing their most powerful and capable image generation model into the Gemini app. It’s called Imagen 4, and it’s a big leap forward. The images are richer with more nuanced colours and fine-grained details, and Imagen 4 is better at text and topography. And with Gemini native image generation, you could edit these images right in the app, there’s a super-fast version of Imagen 4, and it’s 10x faster than the previous model.
Now, Gemini is integrated into Chrome, and that means it’s the personal AI assistant that will be there for us while searching the web, and it understands the context of the page we are on automatically. Gemini will come to Chrome this week in the US for the subscribers.
Veo 3 is available today, and its visual quality is better, and its understanding of physics is better. Veo 3 has native audio generation, which means Veo 3 can generate background sound, dialogues, and sound effects.
Google will now have new subscriptions. Google AI Pro and Google AI Ultra: With the Pro plan, which will be available globally, we’ll get a full suite of AI products with higher rate limits and special features compared to the free version. This includes the Pro version of the Gemini app, previously known as Gemini Advanced, and it starts at 20 USD a month with a 1-month free trial, and there’s the Gemini Ultra. It’s for the pioneers who want cutting-edge AI from Google.
This plan comes with the highest rate limits and the earliest access to new features and products from Google. This plan comes with the highest rate limits and earliest access to new features and products from Google. It’s rolling out in the US today and globally soon, and it’s the VIP pass for Google AI. You will get YouTube Premium and access to 2.5 Pro Deep Think, Flow, and Veo 3, and 30TB of storage compared to the 2 TB you get in the Gemini Pro.
Google Beam at Google IO
Google Beam uses a new state-of-the-art video model to transform 2D video streams into an immersive 3D experience, and behind the scenes, it has an array of six cameras with AI that we can merge all these streams together for near-perfect head-tracking at 60 fps, all in real time.
In collaboration with HP, the first Google Beam device will be available for customers later this year.
Google Meet
Google Meet now includes real-time speech translation to break down the language barrier. It matches the speaker’s tone, expression, and even their gestures. English and Spanish translations are now available for subscribers, with more languages rolling out in the coming few weeks, and real-time translation will be coming to enterprises later this year.
Gmail
With the permission of Gemini models, you can use relevant content across your Google apps in a way that’s private, transparent, and fully in control of the user, and imagine the personalised replies could sound like you—that’s the idea behind the personalised smart replies. Looking at notes and drives and scanning the email for previous reservations, Gemini captures the tone, style, and favourite word choices, then automatically generates a reply. It will be available in Gmail for subscribers later this summer.
AI Mode
For those who want an end-to-end AI search experience, they are introducing a new AI mode. It’s a total reimagining of search with more advanced reasoning. You can ask AI mode longer and more complex queries. Testers have been asking 2-3x longer queries than the traditional search. All of this is available today as a new tab in search AI mode, which will be rolling out to everyone in the US starting from 20th May.
Soon, AI mode will be able to make your responses even more helpful with personalised suggestions based on past searches, and you can also opt in to connect with other Google apps, starting with Gmail. With personalised context, you can see how personalised context makes search ours with recommendations customised just for you, and you can choose to disconnect or connect at any time. Personalised context is coming to AI mode this summer.
They are bringing deep research capabilities into AI mode. It can issue dozens or even hundreds of searches on our behalf. It reasons across all that information to create an expert-level, deep-sighted report in minutes. It includes links throughout so you can easily explore and take action, and a deep search will be coming to AI mode this summer.
We are getting Project Astra’s capabilities, getting to AI mode, and now using the camera search can see and give you helpful information as you go back and forth in real time, and it’s like having a video call with search, and search live comes to AI mode this summer.
We are getting a new try-on feature that will help you virtually try on clothes so you can get a feel for how styles might look on you. To do this, they built a custom image generation model specifically trained for fashion. It enhances 3D shape understanding, which allows them to perceive perceived shapes and depths more accurately, assisting in better visualising the body. Then the AI model will fold, stretch, and drape it on people. Search, help find the price that you want, and buy with the new agentic checkout feature with just one tap. Search security purchases it for you, and it happens under your guidance.
We hope this article was helpful to you. Continue reading GizmoGeek Hub for more tech news, stories, and reviews that will keep you up to date on the newest trends and advancements in technology. We attempt to provide in-depth and unbiased reviews, detailed tutorials, and up-to-date news to help you improve your tech knowledge and expertise.
Keep reading GizmoGeek Hub for more such articles & reviews.
Follow us on YouTube, X (previously Twitter), Instagram, LinkedIn, and our WhatsApp Channel to stay up to date on the latest news, announcements, and behind-the-scenes stuff.
