Faster supply of artificial intelligence models for events

At the Google I/O annual conference on May 9, Sandar Pichai, CEO of Google and Alphabet, announced new developments in the field of artificial intelligence.

Picai said Google is no longer waiting for big events and offers its artificial intelligence models as soon as it is ready.

“We may introduce a new model on a regular day in March or announce important news like Alphaevolve just a week before I/O,” he said. Our goal is to provide the best technologies faster than ever. “

Unnecessary progress of artificial intelligence models

Google’s CEO announced the significant progress of the company’s artificial intelligence models. According to him, the Gemini 2.5 Pro, the latest version of the Jammeh series, has ranked first in all LMARNA platform categories.

This progress is possible thanks to Google’s strong infrastructure, including the seventh generation of TPU chips called Ironwood; A chip that is 5 times faster than the previous generation and its processing power reaches a stunning number of 1.5 excess per unit.

According to Pichai, this processing power has allowed Google to offer faster and cheaper models than competitors: “Not only are we at the forefront of optimized models, but we have also moved the boundaries of the industry.”

Wide welcome of Google’s artificial intelligence

The statistics provided by Pichai show the unprecedented growth of using artificial intelligence around the world:

While Google was processing about $ 5 trillion token last year, it has now reached more than 5 trillion.

More than 5 million developers are building an app using Jammeh; Five times more than last year.

The use of jams on the Vertex AI cloud platform has increased by up to 5 times.

The “Jamna” app now has more than 5 million monthly active users, and the use of the professional version of the 4.3 model in this app has increased by 5 %.

From Research Project to Real Product: Introducing Google Beam

One of the most important introductions of this year’s conference was a product called Google Beam; The commercial version of the Starline Research Project introduced a few years ago. Beam is an advanced video call platform that, using artificial intelligence and six special cameras, transforms the two -dimensional image into a real and real experience.

According to Google, the technology carefully follows the motions of the audience and renders everything at a rate of 2 frames per second. The result is a more natural and closer conversation to the face -to -face meeting. The first Beam devices will be released by HP by the end of this year.

Live and natural translation in Google Meet

Google also announced that it will add live translation of speech to the Google Meet service. This new technology also makes it not only translate, but also imitate the voice of the speaker and even his face states. This feature is currently being tested for English and Spanish users of the Pro and Ultra version and will soon be available to business customers.

“We have entered a new phase of the evolution of artificial intelligence; “The stage in which the result of years of research and development has now been realized to people, businesses and communities.”