- At I/O 2025, Google is likely to announce two new Gemini Pro and Gemini Ultra subscription tiers.
- In addition, the search giant may unveil Veo 3 video generation model, Imagen 4 image generation model, and a larger Gemini Ultra AI model.
- We may also see the Project Mariner AI agent for consumers and the 'Computer Use' AI agent for enterprise customers.
The search giant is gearing up for its annual developer conference, Google I/O 2025, scheduled for May 20 and 21. Much like last year, we may see several new AI announcements related to Gemini, AI agents, Android XR, and more. So in this article, we have compiled the most anticipated announcements and features, likely to be unveiled this year at I/O 2025. On that note, let’s check out the list.
Gemini AI
At the I/O 2025 event, Google is expected to announce two new subscription plans called Gemini Pro and Gemini Ultra. According to TestingCatalog, code strings of Gemini Pro and Ultra have surfaced, and they may offer expanded usage limits for video generation and other capabilities. Currently, Google only offers Gemini Advanced, which costs $20 per month.
Apart from that, it’s also being said that Google might release a larger and more powerful Gemini Ultra model. Google stopped releasing the Ultra series of models after Gemini 1.0 Ultra. Currently, Gemini 2.5 Pro is Google’s most capable AI model, and it’s available to Advanced subscribers and free users as well, but with a rate limit.
Next, at Google I/O 2025, we may see three new AI models: Veo 3, Imagen 4, and Imagen 4 Ultra. Veo 3 is going to be the successor of Veo 2, which is already the best AI video generation model out there. On top of that, Imagen 4 and Imagen 4 Ultra will succeed Imagen 3, Google’s state-of-the-art image generation model.
Google’s AI Agent
At the Google I/O 2025 event, the search giant may finally release an AI agent for consumers. Google did showcase Project Mariner back in December 2024, which can perform actions in the Chrome browser, similar to OpenAI’s Operator AI agent. However, during the announcement, the company said that the early prototype is slow and not always accurate. However, after five months, Project Mariner may be ready to ship.
In fact, some X users have found reference to ‘Computer Use’ which allows “Gemini to use features such as browser access, code execution, and file system capabilities.” It’s also being said that Computer Use is aimed at enterprise customers to run AI agents in virtual desktop instances. Google keeps saying that 2025 is going to kickstart the agentic AI era, so it’s expected that we will see new and powerful AI agents at I/O 2025.
Other Improvements Coming to Gemini
While Google offers a personalized AI model in Gemini which uses your search history to learn about your preferences, at I/O 2025, the company may finally embrace memory and custom instructions. Early leaks show that Gemini is about to receive something called “Personal Context” which is similar to custom instructions on ChatGPT.
Here, you can define what you like, what you don’t prefer, and add your personal information so that Gemini can know you better. Apart from that, memory support is coming to Gemini which will remember crucial parts from your conversations to build a long-term memory about yourself.
Next, Google may allow users to upload files and images while using the Deep Research AI agent on Gemini. And developers will be able to upload folders of their code repositories, or use GitHub integration to easily bring their large codebase to Gemini. Not to mention, Google is working to let users share Gemini Gems, which is similar to custom GPTs on ChatGPT.
Meanwhile, Google is also expected to announce Gemini integration into the Chrome browser. New leaks show that Gemini will appear in the Windows system tray and can be triggered using a keyboard shortcut. Finally, you can install the NotebookLM app on your smartphone, which goes live on May 20.
Android XR
At I/O 2025, Google is expected to preview its Android XR mixed-reality OS to take on Apple Vision Pro and Meta Quest headsets. Samsung has already announced that ‘Project Moohan’, its Android XR headset, is launching this year. Interestingly, Google may announce its smart glasses, powered by the Android XR platform and Gemini AI.
Google’s smart glass looks regular and has a lightweight design, having deep integration with Gemini for a conversational experience. It has vision capabilities and can perform contextual tasks on the go. Basically, the smart glass uses Gemini AI to enhance the user experience in an immersive environment.
Finally, Google has already showcased its bold new ‘Material 3 Expressive‘ design language with Android 16, a week before Google I/O 2025. This is the first time Google has held a separate Android event before the I/O. Google also says that Gemini is coming to WearOS, Android Auto, Google TV, and Android XR platforms.
