The Google I/O 2025 developer conference commenced with an exciting keynote on Tuesday, featuring a range of announcements from CEO Sundar Pichai and other top executives. Attendees learned about a series of advancements in artificial intelligence (AI), including enhancements to the Gemini 2.5 series of AI models, new AI Mode features in Search, the launch of the innovative 3D communication platform Google Beam, and a demonstration of the Android XR platform. For those who couldn’t watch the event live, here’s a comprehensive overview of the key announcements.
Introducing Google Beam
Google’s Project Starline has been rebranded as Google Beam, a platform designed for 3D communications. This technology utilizes an array of six cameras to capture video of users from various angles, allowing an AI system to convert the 2D footage into a 3D light field display. The platform also employs head-tracking sensors to capture users with impressive accuracy at 60 frames per second (fps).
The company has partnered with HP to roll out the first Google Beam devices later this year, with availability limited to select customers initially. Furthermore, additional Google Beam products from various original equipment manufacturers (OEMs) will debut at InfoComm 2025 in June.
Upgrades to Gemini 2.5
The Gemini 2.5 series is set to receive a host of new features, including a Deep Think mode for the 2.5 Pro model, which focuses on enhanced reasoning capabilities now undergoing testing. Additionally, a Native Audio Output feature will be integrated into the Gemini 2.5 models via the Live application programming interface (API), creating more human-like speech generation.
The Gemini 2.5 Flash model will also see significant upgrade, improving its capabilities in reasoning, multimodality, programming, and managing long context interactions, while also reducing costs for developers utilizing the Gemini API, who will gain access to thought summaries and thinking budgets with the latest versions.
AI Mode Enhancements in Search
Another highlight of the keynote was the updates to the AI Mode in Search. Google plans to integrate a custom Gemini 2.5 model to enhance end-to-end AI search functionality. The new Deep Search mode will include a Live Search feature that allows the AI to interact with the device camera and an agentic feature for users to purchase event tickets and book appointments directly through the search interface.
The AI Mode will also introduce additional shopping-oriented features, enabling users to visually search for products, virtually try on a variety of apparel by uploading images of themselves, and utilize AI agents for tracking product prices and facilitating automated purchases. These updates are expected to launch later this year.
Expansion of AI Overviews
The company announced during the keynote that AI Overviews would now be accessible in over 200 countries and more than 40 languages. This expanded AI-powered snapshot of search results will now support additional languages, including Arabic, Chinese, Malay, and Urdu, alongside existing options such as English, Hindi, Indonesian, Japanese, Portuguese, and Spanish.
Android XR Powered by Gemini
Google also showcased a demo of its Gemini-powered Android XR platform, set to serve as the operating system for Samsung’s new Project Moohan smart glasses. Collaborations with other wearable technology partners are also in progress.
These forthcoming smart glasses will incorporate a camera, microphone, speakers, and an integrated display, allowing users to engage in hands-free conversations with Gemini, take photos, control their smartphones, and connect other devices seamlessly.
New Imagen 4 and Veo 3 AI Models
The tech giant unveiled its next generation of image and video generation models: Imagen 4 and Veo 3. Imagen 4 features enhanced text rendering capabilities and improved contextual understanding, alongside boosted image quality and adherence to prompts.
Meanwhile, Veo 3 introduces native audio generation, enabling generated videos to include ambient sounds, background music, and dialogues. Both models are slated for public release later this year.
A new AI-powered filmmaking application named Flow was also introduced. It harnesses the capabilities of Imagen, Veo, and Gemini to create eight-second video clips, which can be combined to form longer scenes, taking prompts from both text and images.
Gemini Integration in Google Chrome
Paid subscribers can now utilize the Gemini AI assistant directly within Google Chrome. A new Gemini button will empower users to summarize web pages or inquire about content, and it will automatically navigate websites based on user instructions. The assistant is equipped to manage activities across multiple tabs simultaneously.
The company has also announced a new AI-enabled tool called Stitch, designed to generate app interfaces from text prompts and templates, including support for wireframes, rough sketches, and existing user interface designs. This tool is currently available for experimentation via Google Labs.
Real-Time Speech Translation for Google Meet
In another update, Google Meet will receive a new AI feature that provides real-time speech translation, aiding users who speak different languages to communicate with minimal delay. The feature currently supports translations between English and Spanish and is available in beta for paid subscribers.
Plans for Google AI and AI Ultra
To round off the announcements, Google introduced its Google AI and Google AI Ultra plans linked to its Gemini features. The Google AI plan will replace the previous Google One AI Premium at a monthly price of $19.99 (approximately Rs. 1,950 in India), while the Google AI Ultra plan will be priced at $249.99 (about Rs. 21,000) per month, providing early access to new features, higher rate limits, and 30TB of cloud storage.
