Google is rolling out Gemini 3 Deep Think mode to its AI Ultra subscribers starting today.
This new capability represents a significant step forward in AI reasoning, enabling the model to explore multiple hypotheses simultaneously before arriving at an answer.
Designed for challenging tasks in math, science, and logic, Deep Think builds on the architecture of previous Gemini 2.5 variants, which have achieved top-tier results in international competitions such as the Math Olympiad.
The model has demonstrated industry-leading performance on rigorous benchmarks, including a score of 41.0% on Humanity’s Last Exam without tools and an impressive 45.1% on ARC-AGI-2 with code execution.
By using parallel reasoning, Gemini 3 Deep Think can tackle problems that typically stump standard models, making it a powerful tool for researchers, developers, and students.
Ultra subscribers can access this feature immediately by selecting “Deep Think” in the Gemini app’s prompt bar.
Key Takeaways:
- Gemini 3 Deep Think mode is now available for AI Ultra subscribers.
- The model employs parallel reasoning to solve complex mathematical and logical problems.
- Benchmarks show leading scores on Humanity’s Last Exam and ARC-AGI-2.
You may also want to check out some of our other recent updates.
Wanna know what’s trending online every day? Subscribe to Vavoza Insider to access the latest business and marketing insights, news, and trends daily with unmatched speed and conciseness! 🗞️
