Google I/O 2025, Google's biggest developer meeting, will be held on Tuesday and Wednesday at Shoreline Amphitheater in Mountain View. We bring you the latest updates from the event to you.
I/O presents product announcements from across Google's portfolio. There's a lot of news about Android, Chrome, Google Search, YouTube, and of course, Google's AI-powered chatbot, Gemini.
Google hosted an independent event exclusively for Android Updates: The Android Show. The company has announced new ways to find lost Android phones and other items, additional device-level features for advanced protection programs, security tools to protect against fraud and theft, and a new design language called Material 3 Expressive.
Everything that was announced at Google I/O 2025.
Gemini Ultra
According to Google, Gemini Ultra (currently only in the US only) provides “the highest level of access” to Google's AI-powered apps and services. At a price of $249.99 a month, it includes Google's VEO 3 video generator, the company's new flow video editing app, and a powerful AI feature called Gemini 2.5 Pro Deep Think Mode.
AI Ultra has high limits on Google's NoteBookLM platform and Whisk. This is the company's image remix app. AI Ultra subscribers can also access Google's Gemini chatbot on Chrome. Several “agent” tools equipped with company project Mariner Tech. YouTube Premium; 30tb storage across Google Drive, Google Photos and Gmail.
Deep Think for Gemini 2.5 Pro
Deep Think is the “enhanced” reasoning mode for Google's flagship Gemini 2.5 Pro model. This allows the model to consider multiple answers to a question before responding, and improve performance with a specific benchmark.
Google didn't go into detail about Deep Think Works, but it could be similar to Openai's O1-Pro and upcoming O3-Pro models. This could potentially use the engine to search and synthesize the best solution for a particular problem.
“Reliable Testers” can use DEEP Think via the Gemini API. Google said it takes additional time to carry out a safety assessment before thinking too deeply.
VEO 3 Video Generation AI Model
Google claims that VEO 3 comes with videos that generate and create effects, background noise, and even dialogue. VEO 3 improves its predecessor, VEO 2, in terms of the quality of the footage that can be generated, says Google.
VEO 3 will be available for subscribers to AI Ultra Plans, which is $249.99 per month, starting Tuesday, on Google's Gemini chatbot app.
Imagen 4 AI Image Generator
According to Google, Imagen 4 is faster – faster than Image 3. And it's quicker soon. In the near future, Google plans to release a variant of Imagen 4, which is up to 10 times faster than Imagen 3.
According to Google, Imagen 4 can render “fine details” like fabric, water droplets and animal fur. It handles both photorealistic and abstract styles, create images in aspect ratio ranges, and create resolutions of up to 2k.
Both Veo 3 and Imagen 4 will be used in Power Flow, the company's AI-driven video tool for film production.
Imagen 4.Image Credit Sample: Google
Gemini app update
Google has announced that the Gemini app has over 400 active users each month.
Gemini Live's camera and screen sharing capabilities will be rolled out this week for all iOS and Android users. Equipped with Project Astra, this feature allows people to have oral conversations with Gemini for almost realistic times, allowing them to stream videos from smartphone cameras and screens to AI models.
According to Google, Gemini Live will begin to integrate more deeply with other apps in the coming weeks. Soon you will be able to provide directions from Google Maps, create events in Google Calendar, and create to-do lists in Google Tasks.
Google says it is updating Deep Research, Gemini's AI agent that generates in-depth research reports by allowing users to upload their own private PDFs and images.
stitch
Stitch is an AI-powered tool that enables people to design front-ends for web and mobile apps by generating the necessary UI elements and code. Stitch will be asked to create an app UI using several words or images and will provide HTML and CSS markup for the design you generate.
Stitching is a little more limited in what it can do compared to several other atmospheric coding products, but there are quite a few customization options.
Google has also expanded access to Jules, an AI agent intended to help developers fix bugs in their code. This tool helps developers understand complex code, create pull requests on GitHub, and handle specific backlog items and programming tasks.
Project Mariner
Project Mariner is Google's experimental AI agent that browses and uses websites. Google has significantly updated how Project Mariner works, saying that agents can take on almost a dozen tasks at a time, and is rolling it out to users.
For example, Project Mariner users can purchase tickets for baseball games and buy groceries online without having to access third-party websites. People simply chat with Google's AI agents, visit websites and take action for them.
Project Astra
Google's low latency, multimodal AI experiences, Project Astra searches for new experiences, Gemini AI apps, and arrays of products from third-party developers.
Project Astra was born from Google Deepmind as a way to showcase near-real-time multimodal AI capabilities. The company says it is currently building Project Astra Glasses with partners including Samsung and Warby Parker, but the company has not yet set a launch date.
Image credit: Google
AI Mode
Google is deploying AI mode. This is an experimental Google search feature that allows US users to ask complex, multipart questions via an AI interface this week.
AI mode supports the use of complex data in sports and finance queries and provides the apparel “Try It On” option. Launching later this summer, Live Live will allow you to ask questions based on what your phone camera is looking at in real time.
Gmail is the first app to be supported in a personalized context.
Beam 3D Teleconference
Previously known as Starline, Beam uses a combination of software and hardware, such as a six-camera array or a custom light field display, allowing users to talk to someone as if they were in the same room. AI models are positioned at different angles and convert video from cameras pointed at the user into 3D rendering.
Google's Beam boasts “nearly perfect” millimeter-level head tracking and 60fps video streaming. When used with Google Meet, Beam offers AI-powered real-time voice translation capabilities that preserve the audio, tone, and representation of the original speaker.
Speaking of Google Meet, Google has announced that Meet has obtained real-time speech translations.
Other AI updates
Google is launching Gemini on Chrome. This will allow people to access new AI browsing assistants, helping them to quickly understand the context of the page and complete tasks.
The Gemma 3N is a model designed to run “smoothly” on phones, laptops and tablets. It will be available in preview from Tuesday. According to Google, it can process audio, text, images and videos.
The company has also announced a large number of AI workspace features that will be featured in Gmail, Google Docs and Google Vids. Most notably, Gmail has got personalized smart replies and new inbox cleaning features. VIDS is getting a new way to create and edit content.
The video overview appeared on NotebookLM, and the company deployed SynthID Detector, a verification portal that helps identify content generated by AI using Google's SynthID watermark technology. Lyria RealTime, an AI model with an experimental music production app, is now available via API.
Please wear OS 6
WEAR OS 6 brings a unified font to the tiles for the appearance of the cleaner app, while the pixel clock gets a dynamic theme that synchronizes the face of the clock and the color of the app.
The core promise of the new design reference platform is to enable developers to build better customizations in their apps along with seamless migrations. The company has released design guidelines for developers along with the Figma design files.
Image credit: Google
Google Play
Google is bolstering its Play Store for Android developers with subscriptions, fresh tools to handle topic pages. Get users to dig into specific interests and sneak peek into app content using audio samples.
The “Topic Browse” page of movies and shows (for now, only us) connects users to apps tied to a bunch of shows and movies. Additionally, developers have a dedicated page for testing and releases, and tools to focus on app deployments and improve them. Developers using Google may also stop releasing live apps if a serious problem occurs.
The subscription management tool has been upgraded with Multi-Product Checkout. Developers will soon be able to offer all subscription add-ons with their main subscription at a single payment.
Android Studio
Android Studio integrates new AI features, such as Journeys, an “agent AI” feature that matches the release of the Gemini 2.5 Pro model. Additionally, “agent mode” can handle more development processes.
Android Studio receives new AI features, including the enhanced “crash insights” feature in the APP Quality Insights panel. This Gemini-equipped improvement analyzes the app's source code to identify potential causes of crashes and suggest fixes.