As it expands the Gemini 3 model lineup, Google has officially introduced the innovative Gemini 3 Deep Think mode. Developed as an enhancement to the Gemini 3 Pro, this mode taps into extra computational power and incorporates an entirely new technical framework. It's specifically tailored to tackle intricate mathematical, scientific, and logical reasoning challenges. At present, this functionality has been seamlessly integrated into the Gemini App, enabling subscribers to Google AI Ultra to activate it alongside the Gemini 3 Pro model. Users can do this simply by choosing 'Deep Think' from the prompt bar. According to official statistics, the Deep Think mode excelled with a remarkable 41% score in the Humanity’s Last Exam test. It also shone in the GPQA Diamond scientific knowledge test, achieving an outstanding 93.8%, and secured 45.1% in the ARC-AGI-2 test. These results place it at the forefront of the industry. The key advancement is its deployment of cutting-edge parallel reasoning technology. This allows the model to simultaneously investigate multiple hypothetical scenarios, ensuring the discovery of the most effective solution.