Gemini gets a new avatar. What does Google's latest play hold for you?

A year after it began its generative AI journey with its natively multimodal model Gemini 1.0, Google has announced the latest version Gemini 2.0, which it calls its "most capable model yet", made for the agentic AI era. ET explains what's new, who can access Gemini 2.0, and where it is being applied and integrated.

How is Gemini 2.0 different?

Google seeks to unlock a new class of agentic experiences with the Gemini 2.0 family. The first model being launched is the experimental Gemini 2.0 Flash.


It has native user interface action capabilities and improvements like multimodal reasoning, long context understanding, complex instruction following and planning, compositional function-calling, native tool use and better latency.

Gemini 2.0 can natively generate images and audio in addition to text. 2.0 Flash can also use third-party apps and services, allowing it to tap into Google Search, execute code, and more. 2.0 Flash is said to outperform Gemini 1.5 Pro on key benchmarks, twice as fast.

In terms of hardware, Gemini 2.0 is powered by Google's sixth-generation tensor processing unit Trillium. Trillium is now being made generally available for customers to build with.

Where will Gemini 2.0 be available?

Developers can now access an experimental release of 2.0 Flash through the Gemini API and Google's AI developer platforms AI Studio and Vertex AI. The new text-to-speech and native image generation capabilities are available only to "early-access partners" with a wider rollout planned in January.

Google said it will expand Gemini 2.0 to its products early next year. A new feature called "Deep Research" will be available in Gemini Advanced. This can act as a research assistant that is able to explore complex topics and compile reports on the user's behalf.