The new upgrades include an AI research assistant with agentic capabilities.
Last year, Google released Gemini, its best effort to tackle OpenAI’s ChatGPT, and yesterday (11 December), the company released Gemini 2.0, its “most capable model yet”.
As part of the announcement, the company said it is releasing an experimental version of Gemini 2.0 Flash, updates to Project Astra, a new Project Mariner, a multimodal live API and Deep Research, a new agentic feature in Gemini Advanced.
Moreover, Google also released Trillium, its sixth-generation tensor processing unit – the company’s custom circuits used to accelerate machine learning workloads – to customers.
According to CEO Sundar Pichai, Gemini 2.0, which is currently in the hands of developers and testers, will enable Google to build new AI agents that bring them closer to their vision of a universal assistant.
What are big the updates?The new Gemini 2.0 Flash builds on and outperforms its previous iteration – the 1.5 Flash – with twice as fast speed, enhanced performance and low latency, Google announced.
The “workhorse model” supports multimodal inputs including images, video and audio as well as multimodal output including natively generated images mixed with text. Moreover, users can also access a chat-optimised version of the 2.0 Flash experimental, which, according to Google, is an “even more helpful Gemini assistant”.
Meanwhile, Google has rolled out Deep Research, its AI research tool in Gemini Advanced on desktop and mobile web browsers, which is expected to be available in a mobile app version in early 2025.
According to Google, Deep Research uses AI to explore complex topics on a user’s behalf. Based on a user’s input, the AI assistant refines its analysis, conducts multiple search queries into its initial findings and provides a “comprehensive report” of its key findings with links to original sources – exportable into a Google Doc.
Deep Research is the first feature in Gemini to bring agentic capabilities – or the ability to act independently – to life, said Google.
In the announcement yesterday, Pichai said: “We’re getting 2.0 into the hands of developers and trusted testers today. And we’re working quickly to get it into our products, leading with Gemini and Search.
“No product has been transformed more by AI than Search,” he said. “As a next step, we’re bringing the advanced reasoning capabilities of Gemini 2.0 to AI Overviews … We started limited testing this week and will be rolling it out more broadly early next year.”
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.