Google I/O 2025: Everything announced at this year’s developer conference

Google I / O 2025The biggest developer conference of the year occurs on the coast at the beach at a mountainee on Tuesday and Wednesday. It is where the event brings you the latest updates.

I / she demonstrates product ads between the Google portfolio. Android, Chrome, Google Search, YouTube and – have certain news about Google’s AI-powered chatbot, the Gemini.

Google has held a separate event dedicated to Android updates: Android show. The company has announced New ways to find lost Android phones and other items, Additional device level features for advanced protection program, Safety tools to protect against scams and theftand Material 3 A new design language called expressive.

This is everything that is announced in Google I / O 2025.

Gemini ultra

According to Google, Gemini Ultra (only in the United States “the highest input in the United States, Google’s AI-Pightsed applications and services” the highest level “. Per month is valuable at $ 249.99 And Google’s VEO 3 video generator includes a powerful AI ability called the company’s new flow video adjustment application and not yet started the Gemini 2.5 Pro deep thinking mode.

AI Ultra comes in a higher level on Google’s notebooklm platform and WhipThe company’s image remixing application. AI Ultra subscribers also allow Google’s Gemini ChatBot in Chrome; Some “agent” tools equipped by the company Project Mariner Technical; YouTube award; Google Drive, Google Images and 30TB Memories in Gmail.

Think in deep twins 2.5 Pro

Deep thinking “Advanced” justification mode Google’s flagship Gemini 2.5 for the Pro model. The model allows you to identify many answers to the questions before answering and increase performance in certain criteria.

Google did not provide detailed information about deep thinking, but it can look like O1-PRO and upcoming O3-Pro models, probably to use an engine to seek and synthesize the best solution.

Deep thinking is possible for “reliable tests” via the Gemini API. Google said he took extra time to conduct security assessments for security assessments before thinking deep.

VEO 3 Video Creating AI Model

Google claims that VEO 3 can create sound effects and even a dialogue to accompany sound effects, background sounds and even videos created. VEO 3, Salafi also improvesVEO 2 can arise in terms of imagery quality, Google says.

VEO 3, Google’s Google’s Gemini chatbot application, Google’s monthly Ultra plan for $ 249.99 to the monthly Ultra Plan, as the text or image can be requested.

Figure 4 EU Image Generator

According to Google, Image 4 fast – Image 3 faster than 3. Soon becomes faster. In the near future, Google plans to release a version of a version that is 10x faster than 3 in IMAGE 3.

Image 4, according to Google, is able to show “delicate details” such as fabrics, water drops and animal furs. Both photorualistic and abstract styles, create pictures in a number of aspects and up to 2K resolution.

Both VEO 3 will be Image 4 used for the flow of powerThe company’s AI-powered video tool is directed to film.

Gemini app updates

Google has announced that Gemini apps have more than 400 active users.

Gemini Live’s Camera and screen sharing capabilities This week will spread to all users on iOS and Android. The project is equipped with Astra, has real-time oral conversations to people with twins, but also leads to video flow from smartphones or from the screen to the AI ​​model.

Google will start gemini Live more deeply in the future with other applications: Google maps will be able to offer instructions and operate with Google Calendar events and Google tasks.

Google updates users to download private PDFs and images, Google, deep research, AI agent of the twins.

Cause to seam

The stitch is a tool to design a web and a mobile app by creating the necessary UI elements and code for people. The stitch is desired Create a few words or even an image with a few words or even an image, to provide HTML and CSS sign for designs.

The stitch is a bit more limited to what a number of other vibe coding products can do, but there are fair amount of customization options.

Google has expanded the entry JulesThe AI ​​agent’s developers aimed to help make mistakes in the code. The tool helps develop developers to understand the complex code, creating inquiries in Github and manage certain rear items and programming tasks.

Project Mariner

Project Mariner is an experimental AI agent that reviews and uses Google’s websites. Google said it was The project has been updated significantly how Mariner worksAllows the agent to take approximately eleven tasks at a time and now spread to users.

For example, project Mariner users can buy tickets to a baseball game or buy online groceries online before the third party website. People can only chat with Google AI agent and visit these sites and take action for them.

Project Astra

Google’s low delay, Multimodal AI experience, Project AstraIn search, the Gemini AI application and third party developers will strengthen a number of new experiences.

The project was born as a way to demonstrate Multimodal AI opportunities, almost real-time, from Google Deepmind. The company now says Samsung and including partners build astra glasses Warby ParkerHowever, the company still does not have the start date.

AI mode

Google AI mode is rolled in experimental Google Search feature It allows people to ask many parts of the United States to users in the United States this week.

AI mode will support the use of complex data in sports and financial surveys and offers the “test” options for clothing. This summer will allow you to ask questions, based on what happened later in your phone’s camera.

Gmail is the first application to be supported by an individual context.

Beam 3d teleconference

Beam is called the star line beforeUses a program and hardware combination, including six-camera and special light area screen to allow one user to speak with someone as in the same session room. The AI ​​model changes the video to the video, placed in different aspects and aimed at the user, and aimed at 3D.

Google’s beams have a “perfect perfect” millimeter-level head followed and 60fps video stream. When used with a Google meeting, the beam provides a real-time speaking feature that works with AI, which protects the sound, tone and phrases of the original speaker.

The talking of the Google meeting, Google has announced that It takes a real-time speech translation to meet.

More AI updates

Google begins Twins on ChromeIt will allow people to quickly understand the context of a page and help a new AI browsing assistant that will help you quickly understand the tasks.

Gemma is a model designed for 3n Run “smoothly” on phones, laptops and tablets. There is preview starting on Tuesday; According to Google, you can manage audio, text, pictures and videos.

The company also announced a ton AI Workspace Features Gmail comes to Google Documents and Google Vids. Basically, gmail, hidden smart answers and a new box cleaner, get new ways to create and edit content while vids.

Video views come to notebooklmand the company has spread Detector in the syntAI Synthidic watermarking technology to determine the emerging content of the AI, a verification portal that uses Synthid watermarking technology. Lyria RealtimeAI model, which strengthens the application of experimental music production, is now available through an API.

Wear OS 6

Wear OS 6 Brings a single font for a clean application view and gets a dynamic themed that syncs the app colors with pixel watch tracking faces.

The basic promise of the new design reference platform is to allow you to better customize in applications along with seamless transitions. The company is released A design instruction for developers in conjunction with Figma design files.

Play Google

Google has Play Store Up for Android Developers Fresh tools to manage subscriptions, users are aware of sound samples and sound patterns to make sound samples and a new checking experience to sell to the content content of the application and sound examples.

“See topics for movies and shows” will be connected to the applications for tons of shows and movies.

Subscription management tools also get a great verification. Devs will soon be able to offer subscription additives with basic subscribers, all in one payment.

Android studio

Android Studio combines new AI featuresAn agentic AI capacity, which is coincident with the release of the Gemini 2.5 Pro model, including “travels”. And “Agent mode” will be able to manage more complex development processes.

Android studio will receive new AI opportunities, including an advanced “accident concepts” feature in the quality concept panel. Equipped by Gemini, this improvement will analyze the source code of an application to determine the potential reasons for accidents and offering adjustments.