Google Officially Introduces the Gemini 3 Deep Think as its Most Advanced Reasoning Mode: Now Available to Ultra Users on the Gemini App

The California-based tech giant – Google has now officially introduced the Gemini 3 Deep Think mode. This is claimed to be the brand’s most advanced reasoning mode yet, and is currently accessible by Ultra users.

Check out more about it and its features below.

Gemini 3 Deep Think – Now Made Available to Google’s AI Ultra Users

The latest Gemini 3 Deep Think mode has been build on top of the Gemini 2.5 Deep Think variants, and it makes use of advanced parallel reasoning in order to explore many hypothesis at the same time, and provide a sophisticated output. Meaningful improvements have thus been brought to its reasoning capabilities, and it is designed in such a way to easily tackle complex math, science, and logic problems.

In terms of its performance, it has achieved top scores on the industry’s leading rigorous benchmarks like Humanity’s Last Exam and ARC-AGI-2. On the Humanity’s Last Exam, it managed to score 41.0% without the use of any tools. With code execution, the Gemini 3 Deep Think attained a 45.1% score on ARC-AGI-2.

As mentioned, Google has made its Gemini 3 Deep Think mode available to Ultra users, and they can access it on the Gemini app by selecting the ‘Deep Think’ mode in the prompt bar, and Gemini 3 Pro in the drop-down for models.

In related news, the Google Workspace Studio was also introduced recently, and it offers users the capability to create AI agents in just a few minutes. It is set to be rolled out over the next few weeks, initially for Enterprise users, and later for other paid users.

Stay tuned for more updates.