Google has introduced Gemini 2.5 Deep Think, its most advanced AI reasoning model, capable of processing multiple ideas at once to provide the best solution.
This new model runs multiple agents in parallel, using computational resources to explore various possibilities before selecting the most accurate answer.
Unveiled in May 2025 at Google I/O, Gemini 2.5 Deep Think is the first publicly available multi-agent AI from the company.
It recently helped Google win a gold medal at the International Math Olympiad (IMO), where its version solved complex mathematical problems.
For researchers, scientists, and academics tackling hard problems: Gemini 2.5 Deep Think is here. 🤯
It doesn’t just answer, it brainstorms using parallel thinking and reinforcement learning techniques. We put it into the hands of mathematicians who explored what it can do ↓ pic.twitter.com/lsBq3ICvuM
— Google DeepMind (@GoogleDeepMind) August 1, 2025
Available to Google’s Ultra plan subscribers, priced at $250 per month, Gemini 2.5 Deep Think is housed in the Gemini app.
Google is now sharing its IMO-winning model with a select group of academics and researchers. Unlike traditional models, this system can take hours to reason, making it invaluable for academic use cases, as reported by Tech Crunch.
Gemini 2.5 Deep Think also performed impressively on the Humanity’s Last Exam (HLE), scoring 34.8%, surpassing competitors like xAI’s Grok 4 and OpenAI’s o3.
The AI integrates with tools like code execution and Google Search, providing more detailed and creative responses than most models. It has shown potential to transform fields like research and development.
While multi-agent systems are more resource-intensive than traditional models, Google has made the Gemini 2.5 Deep Think available through its premium subscription to fund the necessary infrastructure.
Google plans to roll out Gemini 2.5 Deep Think to select testers via the Gemini API to understand its use in business and development, further solidifying its AI leadership.