I/O 2025 announced Google: AI mode of search, 3D video calls, Android smart glass, and more |

Google’s annual developer conference, The I/o 2025Announcements, demo, and further looking promises arrived with their own normal hurry. This year’s I/O continued the company’s all-in approach on Artificial Intelligence with significant updates in its product lineup, held at the noise amphitheater at Mountain View. A $ 250/month from an eye-popping was re-discovered with AI from Ultra subscription, Google Gave a smuggusboard of announcements, which was really related to the impressive to your wallet. Come, let go down a run down by Google which Google has announced in I/O 2025.

Gemini 2.5 models also become smart

Gemini of Google, AI modeE Lively upgraded in its lineup, providing strength to its many services. The new Gemini 2.5 Pro is now a “enhanced logic mode” called deep think, which allows the model to consider several hypotheses before responding – especially useful for complex mathematics and coding problems. According to Google, it holds it on top of the LMARENA leaderboard in all categories, with the first generation of Gemini Pro since there were more than 300 points.Meanwhile, Gemini 2.5 flash – more efficient, cost -effective version – logic, multimodulity, code and long reference capabilities have been improved. Google claims that it now ranks second for only 2.5 Pro on benchmark tests, while reactions require 20–30% less tokens. For developers in search of powerful, budget-friendly AI tools, this is equivalent to receiving premium performance at the mid-range price point.

Project Starline Real and a new name, google beam gets

Remember the Futuristic 3D video conferencing booth of Project Starlline, Google, who promised to do so that it seems that your conversation partner was sitting right from across you? This is no longer a research project – it is becoming a real product called Google Beam.The technique uses a six-camera array and AI to merge video streams in a realistic 3D experience seen on the performance of a light field. Google claims that “near the right head tracking, below millimeters” in 60 frames, all are happening in real time. HP will be the first to release the beam device for enterprise customers earlier this year, with companies like Deloite, Duoolingo and Celsforce already signed up.I am not saying that it is going to change the zoom throughout the night, but it can actually make those endless virtual meetings a little less soul-tail.

AI Ultra Membership Service can be that no one asked for, but they want it

In the category “Things No No No No No No No No No No,”, Google announced a new premium subscription tier called AI Ultra, priced at a jaw $ 250 per month. Yes, you read correctly, two hundred and fifty dollars. Monthly.For this Netflix-annual-sabotent-time-twenty-twenty price tag, customers have early access to unlimited use of facilities such as Google’s latest AI Tool and Deep Research. The package includes 30TB storage in access to experimental features such as Google Photo, Drive and Gmail, Plus YouTube premium and project merge. New customers can get a discount of 50% for the first three months, which still descends $ 125 monthly.Google clearly believes that some people are willing to pay premium prices for state -of -the -art AI tools. The rest of us will just have to wait for these features so that eventually fall down to more affordable levels.

Google Search gets a complete change with AI mode

Google said one of the AI ​​overview “one of the most successful launch in search in the last decade”, the company is going to change how we find the web with AI mode. In Google Search This separate tab allows users to ask more complex questions and follow with additional questions – essentially transforms Google search into a interactive AI experience.Initial examiner is asking from questions that are two to three times longer than traditional discoveries, suggests that users are already compatible with this new paradigm. The AI ​​Mode is rolling for everyone in the US starting this week, operated by Gemini 2.5.Google is also adding features such as deep discovery, which extends the query of the background to hundreds to create extensive search reactions, and an AI purchasing experience that helps users find items and even virtually virtually tries on clothes “by uploading a picture of themselves” on clothes “.

Google brings smart glasses from shade with Android XR

After last week Android showWhere Google previewed some Android 16 features, I/O provided more information on Google’s mixed reality plans. The company unveiled it Android XR Stoning for enhanced, mixed and virtual reality tools.The most interesting development project is Aura, which is a prototype of Android XR-operated smart glass developed with xarel. These glasses will include Gemini integration and a large area of ​​visual as well as built-in cameras and microphones. Google is partnered with popular eyewear brands including Samsung, gentle monster and warbee parker to create a more stylish option.Android XR will support live translation, directional navigation through a mini Google maps display and the ability to watch the 360-degree video. Following the efforts of Google Glass and previous AR, it seems that the company’s most severe push is still in wearable computing.

Gmail gets AI-managed smart answers that really look like you

Google’s AI is coming to your inbox, but in a useful way. Gmail will soon facilitate individual smart answers that analyze your writing style and previous email so that you sound like -like you. The system considers your specific greetings, tones and even preferred word options to generate more relevant answers.With your permission, Gemini can draw information from both your Gmail and Google Drive to craft these reactions. For example, if a friend is asking about the road trip taken by you, Gemini may refer to your previous journey programs stored in Docs to suggest a detailed response.This feature will be launched for English users on web, iOS and Android platforms through Google Labs in July. It is either incredibly convenient or slightly unnecessary, it depends on how you feel about mimicking your writing style.

Google Meat Real-Time AI adds translation

Video calls with international colleagues or friends are going to be very easy. Google Meet is adding an AI-operated real-time translation feature that converts your speech almost immediately into your conversation partner’s favorite language. The system also tries to match your tone and rhythm when translating.This feature initially supports the translation between English and Spanish, in which more languages ​​are planned for the near future. It is rolling Google AI Pro and Ultra Subscribers first in beta. It can be a game-chain for global businesses and multilingual families.

Google’s AI agent is becoming more powerful

Google’s experimental AI agent, Project Meriner gained significant upgradation. The system can now handle 10 different tasks simultaneously, from booking flights to research on topics to compare shopping options.The latest version also introduces a “teach and repeat” function, where you can perform a task once, and learns how to do similar tasks in the future. Google is bringing the capabilities of the project mergeer for developers through Gemini API, in which automation with reliable examiners such as anywhere and Unepaths are already constructing it with it.

Google’s vision for a universal AI assistant takes shape with the project Estra

Remember when AI assistant was glorified just weather forecasts and timer-seters? The goal of Project Estra, Google’s ambitious universal AI auxiliary prototype, is to feel those days like ancient history. In I/O 2025, Google displayed ASTRA’s increased capabilities, showing how it can now use your phone camera, to “see” the world around your nearby world and take action without clear commands. Google is positioning Astra as the culmination of its AI supporting task, which is capable of diving into his email to find bike glasses to research and even call local shops on their behalf. Although some people may feel great about giving AI access to their digital life, productivity benefits may be enough, if you are comfortable with AI auxiliary who can sometimes pipe with unwanted advice, while you are trying to fix your bike chain peacefully. Estra features are already seen in Mithun Live, which is now available to all Android users and is rolling for iOS starting today.

Google’s VO and Imagene Gets Real (Er)

If you thought that artificially produced video comic, you should have seen Google’s keynote speaker, as it was previewed by the new model. The latest version of VEO 3, Google’s AI video generator, now supports audio generation – adds dialogue to ambient sounds, music, and even AI -borne video. The previous versions could only create silent clips, which was not particularly useful in a world where Tikkok and YouTube dominated.Google’s text-to-image model has been improved to generate more photorialistic images with better handling of excellent details such as imagene 4, clothing, water drops and animal fur. It can now export images in various formats and resolutions up to 2K.Google is also offering flow, an AI film production app that makes these techniques. Flow allows users to create an eight-second AI-related video clip which, depending on the text signal or images, to stitch the clip together for long videos with visual-bedr tools. Think of it as a mini film production studio in your pocket – although one that sometimes makes a strange interpretation of your creative vision.

Stitch is an AI-Janit UI designer

For developers and designers, Google unveiled stitch, an AI-powered equipment that produces user interfaces based on text details. Users can provide screenshots of wireframe, rough sketches, or other designs to direct stitch’s output, making the prototype app and website design quickly easier.Although it has more limited capabilities than some competitive AI design tools, stitch provides HTML and CSS markup for designs that generate it, which helps to brid down the gap between the concept and implementation. The experiment is currently available through Google Labs.