Google is releasing the second version of its AI model, Gemini 2.0. It is supposed to be twice as fast as its predecessor. The model will also control virtual assistants that assist users.
Gemini 2.0 is initially being rolled out in an experimental version called Gemini 2.0 Flash, which is only available to developers. The AI model can generate text, images, and audio in multiple languages. It can also provide assistance with Google searches or coding.
What’s interesting about the new version is that Gemini now also lets you build ‘agents’ or ‘actors’. ‘We’ve invested the last year in developing more ‘agent-like’ models. That means they can understand more about the world around them and think multiple steps ahead. They can also take actions on your behalf, without your supervision,’ says Sundar Pichai, CEO of parent company Alphabet, in a press release. For him, the new version of Gemini heralds ‘a new era’ of ‘agentic’ generative AI.
The model will be available to the general public in early 2025. Google plans to implement the model, largely via AI agents, in its various products, including its search engine, by then.
For example, the company is currently testing a web assistant that helps with online shopping, called Mariner. Also in the pipeline is ‘Jules’, which should help programmers remove bugs from software code. The company is also working with game developer Supercell on an AI assistant for video games.