A new AI model for the agency

Note from Google and CEO Alphabet Sundar Pichai:

Information is the basis of human progress. That is why we have been focusing for over 26 years on our mission to organize information in the world and make it available and useful. And that's why we still push the limits of artificial intelligence to organize this information at each input and make it available with any exit so that it can be really useful for you.

It was our vision when In December last year we presented Gemini 1.0. The first model built as natively multimodal, Gemini 1.0 and 1.5 has led great progress with multiplicity and a long context to understand information in the text, video, images, audio and code and process much more.

Now millions of programmers are building from the twins. It helps us imagine all our products – including all 7 out of 2 billion users – and create new ones. Notebook It is a great example of what multimodality and long context can enable people and why he is loved by many.

Over the past year, we have invest in developing more agency models, which means that they can understand more about the world around you, think about many steps and take action on your behalf, with your supervision.

Today we are glad that we can launch our next era of models built for this new agency: introduction of Gemini 2.0, our most talented model. Thanks to the new multimodality achievements – such as the native image and audio output – and native tools, will allow us to build new AI agents who bring us closer to our vision of a universal assistant.

Today we get 2.0 in the hands of programmers and trusted testers. We quickly work to introduce it to our products, leading from Gemini and Search. From today, our experimental model Bligini 2.0 Flash will be available to all Gemini users. We also launch a new function called Deep researchwhich uses advanced possibilities of reasoning and a long context to act as a research assistant, study of complex topics and compiles reports on your behalf. It is available in a twin advanced today and you can Learn more on our website.

No product has been transformed more by artificial intelligence than searching. Our review of artificial intelligence now reach 1 billion people, enabling them to ask completely new types of questions – quickly becoming one of our most popular search functions in history. In the next step, we introduce advanced Gemini 2.0 reasoning for AI reviews to solve more complex topics and multi -stage questions, including advanced mathematical equations, multimodal questions and coding. This week we started limited tests and we will develop them more broadly at the beginning of next year. Over the following year, we will continue to introduce AI to more countries and languages.

Progress 2.0 is based on ten -year investments in our diverse full approach to AI innovation. It is based on non -standard equipment, such as Trillium, our sixth generation of TPU. TPU fed 100% of the Gemini 2.0 training and inference, and today is Trillium generally available To customers, so that they can also build with it.

If Gemini 1.0 concerned the organization and understanding of information, Gemini 2.0 is that it is much more useful. I can't wait to see what the next era will bring.

-Sundar


LEAVE A REPLY

Please enter your comment!
Please enter your name here