Google again showed its latest innovation during the annual developer conference I/O 2025 on YouTube. This year, the focus was clearly on artificial intelligence and how Gemini models integrate deeper into the Environmental Google system. Let’s go to the most important news from this year.
1. Google Ai Ultra-A Premium Ai Subscribe
Google presented Google has superior A subscription that brings together the best in Amnesty International in a package. This subscription allows users to access the most advanced features and models in the AI’s ecosystem of Google. This is very frustrating as Google seems to track Openai’s strategy by subscribing to Super Duper Alame Premium at $ 250 per month, which itself contains additional services such as YouTube Premium, storing a 30 TB cloud and new flow service.
2. Deep thinking is a new experimental thinking method for Gemini 2.5 Pro
Gemini 2.5 Pro has now been improvedDeep thought– It is an enhanced experimental reason for math and coding. At the same time, Gemini 2.5 Flash gets increased efficiency and new features. This situation uses the latest Google research to think and think, including parallel thinking techniques.
Deep thought: It can question their assumptions and early conclusions, allow longer thinking processes to deal with the most complex problems, and improve the solution through repeated reviews.
3. Google Beam – the next generation video connection
After several years of developing Project Starline, I have now provided Google Google RayAI-moving video contact platform. Technology uses a new video model to convert 2D video flows into realistic 3D experiments using six cameras and AI. Beam uses almost an ideal majority, up to a millimeter level, and works on 60 frames per second in actual time. The result is a very natural and overwhelming conversation experience. The first Google Beam units will be available to the first customers later this year in cooperation with HP.
4. Astra Bir Gemini Live Project
The feature is already available to all Android users and is now launched for iOS users.
The ASTRA project, which was first displayed at the previous I/O conferences, is now integrated into Gemini Live. This feature explores the future opportunities for a global AI assistant who can understand the world around you.
Gemini Live now includes camera sharing and screen sharing functions from Project Astra. Users have already found interesting uses, from interview preparations to marathon training. The feature is already available to all Android users and is now launched for iOS users.
5. The agent’s position – AI behaves on your behalf
Google presented the agent based on the research project Mariner Project. These are the systems that combine the intelligence of advanced artificial intelligence models with access to tools, so that they can conduct procedures on behalf of the user and under their control.
Project Mariner is an initial model for early research with computer use functions to interact with the web. Google made great progress with new multi -task features and a method called “Teach and Repier”, where you can show a task once and learn plans for similar tasks in the future.
6. I Google Search
Google introduces a new brand Artificial intelligence modeIt is a complete review of the search experience. With more advanced thinking, users can ask longer and more complex questions. The first laboratories asked two to three times longer questions than traditional searches.
Artificial intelligence mode is available as a new tab directly in research and is now available to everyone in the United States. Google claims that the artificial intelligence response maintains the quality and accuracy that users expect from research, which is the fastest in industry. As of this week, Gemini 2.5 also comes to search in the United States.
7. Customization with personal context
Google makes artificial intelligence more useful by customizing using something they call personal context. With the user condition, Gemini can use the relevant personal context on Google applications in a special, transparent and fully varied in the user’s title.
An example of this is a new personality smart responses in Gmail. If a friend sends an email and asks for advice for a wild trip you have previously made, Gemini can search for Google Drive emails and previously for a response to a specific response. It matches your typical greetings and picks up a tone, style and even choosing the favorite words.
8. The flow of a new tool for directors
Google presented her latest video models I see 3 That is now built -in sound generation. They also presented Imagen 4, the latest and latest image generation model. Both are available in applying Gemini and opening new opportunities for creativity.
flow Google’s new Acting Film Tools are especially designed for VEO, Imagen and Gemini. The tool aims to help creators and directors to explore ideas without restrictions and create clips and cinematic scenes. Flow provides users with the opportunity to create fixed content using elements created from artificial intelligence, camera controls and non -welded stage. The tool is available to subscribers by Google Ai Pro and Ultra in the United States, with plans to expand in more countries.
Swedish perspective
For Swedish users, many of these features will become available during the year. Indeed, we can see integration with Swedish companies such as Volvo Cars, as Goeni AI’s assistant from Google will become a natural part of the auto system. In addition, Google announces that an artificial intelligence overview of Sweden’s research will be launched, giving users more intelligent search results and more personal research.
