Google I/O 2025 developer conference’s keynote on Tuesday was a packed one. During the session, company CEO Sundar Pichai and other executives announced a plethora of new artificial intelligence (AI) updates and features. Some of these include new capabilities in the Gemini 2.5 series of AI models, updates to AI Mode in Search, expansion of AI Overviews, introduction of the new 3D communication platform Google Beam, and a demonstration of the Android XR platform. In case you did not catch the event live, here’s a quick roundup of everything that was announced.
Google Beam
The tech giant’s Project Starline is now being introduced as Google Beam, a 3D communications platform. It uses an array of six cameras to capture a video stream of the user from different angles. Then, an AI system combines them to turn the 2D feed into a 3D light field display. The company is also using head-tracking sensors to accurately capture the user at 60 frames per second (fps).
Google is working with HP to introduce the first Google Beam devices later this year. The initial devices will only be provided to select customers. Additionally, Google Beam products from original equipment manufacturers (OEMs) will be made available via InfoComm 2025, which is set for June.
Gemini 2.5 Upgrades
Gemini 2.5 series is also getting several new capabilities. A new Deep Think mode is being added to the 2.5 Pro model, which is being called an enhanced reasoning mode. The feature is currently under testing. Native Audio Output, an expressive and human-like speech generation capability is also being added to Gemini 2.5 models via the Live application programming interface (API).
Google is also updating the Gemini 2.5 Flash model with improved capabilities across reasoning, multimodality, code and long context. The model will also be more cost-effective to use. Developers using the Gemini API will also get thought summaries and thinking budget with the latest models.
AI Mode in Search
Another major talking point from the keynote session was the AI Mode in Search. Google is now planning to power the end-to-end AI search with a custom Gemini 2.5 model. The AI Mode is also getting a new Deep Search mode, a Live Search feature that lets the AI tool access the camera of a device, and a new agentic feature to let users purchase event tickets and book an appointment directly from the interface.
AI Mode in Search is also getting new shopping-focused features. Users will now get to visually search for the product they want, try out a wide selection of apparel virtually just by uploading a picture of themselves, and use AI agents to track prices of products and make purchases automatically. These features will be added later this year.
AI Overviews Expansion
The Mountain View-based tech giant announced the expansion of AI Overviews during the keynote. The AI-powered search result snapshot feature will now be available in more than 200 countries and over 40 languages. With this update, it will support Arabic, Chinese, Malay, and Urdu, which join existing language options such as English, Hindi, Indonesian, Japanese, Portuguese, and Spanish.
Gemini-Powered Android XR
During the keynote session, Google also showcased a demo of its new Gemini-powered Android XR platform. It will be the operating system for Samsung’s upcoming Project Moohan smart glasses. The company is also working with other wearable partners.
These Android XR smart glasses will feature a camera, microphone, speakers, and a display screen over the glasses. Users will be able to have hands-free conversations with Gemini, ask it to capture images, control their smartphone and other connected devices, and more.
Imagen 4 and Veo 3 AI Models
Google also unveiled the next generation of its image generation model, Imagen 4 and video generation model Veo 3. Imagen 4 now comes with improved text rendering and contextual understanding of text placement, as well as improved image quality and prompt adherence.
With Veo 3, the company is adding native audio generation capability, which means the generated videos will now feature ambient sounds, background music, as well as dialogues. Both of these models will be released to the public later this year.
The company is also launching a new AI-powered filmmaking app dubbed Flow. It leverages Imagen, Veo, and Gemini to generate eight-second-long video clips. Multiple clips can also be stitched together to create a longer scene. The app accepts both text and images as prompts.
Gemini in Google Chrome
Paid subscribers will now be able to access the Gemini AI assistant within Google Chrome. A new Gemini button will let users summarise a web page or ask questions about the content. It can also navigate websites automatically, based on user prompts. The AI assistant can also work on multiple tabs at the same time.
Google also unveiled a new AI-powered tool that can generate app interfaces based on text prompts and templates. Dubbed Stitch, the app also supports wireframes, rough sketches, and screenshots of other user interface (UI) designs. It is currently available as an experiment via Google Labs.
Speech Translation in Google Meet
The company is also adding a new AI feature to Google Meet. The video conferencing platform will now support real-time speech translation and can help speakers with different native languages converse with minor lag. Currently, the feature can translate English and Spanish. It is currently available to paid subscribers in beta.
Google AI and AI Ultra Plans
Finally, the tech giant also introduced the Google AI and Google AI Ultra plans for its suite of Gemini features. The former replaces the Google One AI Premium plan and will be available for $19.99 (Rs. 1,950 in India) per month, while the Google AI Ultra plan will cost $249.99 (roughly Rs. 21,000) a month. The latter will get all the new features first, offer higher rate limits, and provide 30TB of cloud storage.
