Google is reportedly processing a rollout of Gemini 3 Deep Think mode to Google AI Ultra subscribers in the Gemini app to enhance reasoning capabilities for complex math, science, and logic problems that challenge advanced models.
Considering handling intricate tasks, this mode provides a substantial boost. It excels on demanding benchmarks, achieving 41.0 per cent on Humanity’s Last Exam without tools and an unprecedented 45.1 per cent on ARC-AGI-2 with code execution. These results position Gemini 3 Deep Think as a leader in the industry for rigorous evaluations.
The technology is based on sophisticated parallel reasoning, enabling the simultaneous investigation of various hypotheses. This method is a direct extension of the capabilities found in the Gemini 2.5 Deep Think variants.
News: Delta Government Unveils ₦3bn Plan to Rebuild State Media, Modernise Public Communication in 2026
The current version stands as the top-performing AI model on the ARC-AGI-2 reasoning benchmark.
In a blog published by the company, Google clarified that this achievement is due to its utilization of advanced parallel reasoning, which allows it to “explore multiple hypotheses simultaneously.”
The previous versions have recently achieved gold-medal status at both the International Mathematical Olympiad and the International Collegiate Programming Contest World Finals, showcasing exceptional performance in competitive mathematical and programming events.
Ultra subscribers receive instant access to the Gemini 3 Deep Think mode. They can activate this feature by selecting “Deep Think” in the prompt bar and choosing Gemini 3 Pro from the model dropdown menu within the Gemini application.
