Google Launches Deep Think AI Tool, Elevating Problem-Solving Capabilities
Google unveils Deep Think, an advanced AI tool in the Gemini app with superior problem-solving capabilities.
Key Points
- • Deep Think launched in the Gemini app for AI Ultra subscribers at $250/month.
- • It scored 34.8% on the Humanity’s Last Exam, outperforming competitors.
- • Deep Think employs parallel thinking for solving complex problems and was praised by the IMO president.
- • The tool is available to selected academic researchers for enhanced studies.
Google has officially launched its advanced AI tool, Deep Think, in the Gemini app on August 1, 2025. This new feature is part of the AI Ultra subscription service, which costs $250 per month. Deep Think is tailored for solving complex mathematical and coding challenges, distinguished by its use of parallel thinking techniques to explore multiple solutions simultaneously, potentially revolutionizing the approach to problem-solving in academic and practical contexts.
Deep Think’s impressive capabilities were demonstrated in an academic benchmarking test known as Humanity’s Last Exam (HLE), where it achieved a score of 34.8%. This performance outstripped that of its competitors, including xAI's Grok 4 with 25.4% and OpenAI's o3 at 20.3%. Furthermore, Deep Think is recognized for its extraordinary performance at the International Mathematical Olympiad (IMO), where it achieved a gold-medal standard. The president of the IMO praised the AI's solutions as 'astonishing in many respects.'
This tool is not just an incremental upgrade; it represents what Google describes as its most advanced AI to date, surpassing earlier iterations like Gemini 2.5 Pro. Selected academic researchers have been granted access to Deep Think to facilitate their studies, underscoring Google's intent to integrate this technology into scholarly research. Users wishing to access Deep Think can activate it in the Gemini app for Android, where it seamlessly collaborates with other Google tools for an enhanced user experience.
Despite these advancements, some skepticism remains on the practical integration of such sophisticated AI in everyday problem-solving scenarios. As experts assess its capabilities, the focus will be on how effectively Deep Think can bring about innovation across various fields.
As Google continues to roll out Deep Think and expand access to certified testers, its long-term impact on AI development and application remains to be seen, setting the stage for future competitions in the AI landscape.