Google has made its powerful Deep Think AI tool available in the Gemini app to Google AI Ultra subscribers. The new release incorporates feedback from early testers and research breakthroughs, offering a significant improvement over its previous version.
Deep Think is designed to assist users in solving complex problems by utilizing parallel thinking techniques. It allows Gemini to generate multiple ideas simultaneously and refine them over time, enabling users to arrive at creative solutions. This tool can be particularly useful for tasks such as iterative development, scientific discovery, algorithmic development, and coding.
The AI model’s performance has been tested on various benchmarks, achieving state-of-the-art results in areas such as competitive code performance and expertise in science and math. However, it’s essential to note that the model still requires careful evaluation of safety and responsibility measures to ensure its usage is safe and beneficial.
To use Deep Think in the Gemini app, simply toggle “Deep Think” when selecting 2.5 Pro in the model dropdown, and explore its capabilities with tools such as code execution and Google Search. Further updates on the tool’s usability for developer and enterprise users are planned for release in the coming weeks.
Source: https://blog.google/products/gemini/gemini-2-5-deep-think