Google I/O: More personalized (and useful) artificial intelligence to get Gemini and applications – computers
Google I/O In contact is created by 3D images in the film film produced by the artificial intelligence produced by the VEO model. And the creative ability to combine rainbows with animals, balloons and gums that exhibit the artistic variation of the AI model is impressive. It is Google’s headquarters, with the performance of the technology giant CEO Sundar Pichai, goes to the right of the mountain view.
From the last I/O, Google has launched more than 20 versions of its AI models, he said. Gemini 2.5 Pro is a highlight model. Gemini Blue came at the end of Pokemon, API’s “affectionate nickname”: artificial Pokemon Intelligence.
Sundar Pichai said he had filed a case against 9.7 billion tokens last year and that this year is already in 480 billion tokens this yearAn increase of 50 times. The exploration has also increased, with more than 1.5 billion surveys by manufacturing AI.
One of the technologies submitted is Google Beam, which is a system that can use different 2D rooms to create a 3D picture User, in real time. This system will be displayed in a few months.
Google translation also develops, maintains a liquid conversation between different languages, which is available on Google Meet, Between English and Spanish for service subscribers at the moment. Gemini Live is now available on iOS and Andro smartphonesD.
When it comes to the project mariner, Google proposes a model with sophisticated tools that google can carry out the user’s work. Multi -tasking capacity now allows you to have 10 simultaneous applications. The system learns to use the user and soon Gemini reaches API. The AI agent communication system, agents 2 agents, and dozens of partners are already preparing their APIs. The abilities of AI agents begin to reach Chrome, search and Gemi.
Oh Intelligent Gmail Automatic Response Customization System, Signs ensure that the user is similar to the user. This is because the system usually analyzes the process of writing in previous emails, according to the voice and style and user. The system reaches Gmail in the summer for payment subscribers.
Gemini with the latest progress in Google’s AI model comes in early June, Demis Hasabis from Google Deepmind. The system has the ability to debit different tones in 24 languages, as an alternative between languages, maintaining a single voice. Gemini API already allows you to test this system.
Gemini penetration can produce five times more models than Gemini Flash 2.0 Light. The system reaches the Gemini 2.5 flash, which can produce a code more faster for those who need programming help. The Deep Think Mode Google’s AI model helps to give more reflective answers and it has already begun to get good results in benchmark tools.
(In update)
