Google has unveiled the next-generation version of its AI-powered virtual assistant Gemini, which it says is faster and smarter than existing models.
The US tech giant said Gemini 2.0 Flash, the first model in the update, is able not only to understand a mixture of text, audio, video and image inputs – known as multimodality – but is also capable of producing multimodal output, such as images mixed with text, for the first time.
Google said 2.0 Flash is also able to “natively call” additional tools such as Google Search to help with queries, as well as link to other third-party functions.
In addition, Google said Gemini 2.0 has been built with the concept of AI agents in mind, which the company said is likely to be the next stage of generative AI development.
AI agents are smaller, specifically made versions of AI models designed to be experts in one specific area or topic – for example one trained to be a travel planner or another to help software developers write code.
Alongside Gemini 2.0, Google said it is launching a number of prototype AI agents for users to try.
The announcement came on the same day Apple began the rollout of its own suite of AI tools, known as Apple Intelligence, bringing the features to the UK for the first time, as the arms race to supply consumers with the most appealing AI tools continues among the tech giants.
Sundar Pichai, Google’s chief executive, said: “Today we’re excited to launch our next era of models built for this agentic era: introducing Gemini 2.0, our most capable model yet.
“With new advances in multimodality – like native image and audio output – and native tool use, it will enable us to build new AI agents that bring us closer to our vision of a universal assistant.”
Mr Pichai confirmed that Gemini 2.0 would be steadily rolled out to developers and more general Gemini users between now and January but a chat-optimised version would become available immediately for Gemini users on the web and the dedicated app.