In May 2025, Google introduced Gemini 2.5 Deep Think at the Google I/O conference. This advanced AI reasoning model features a multi-agent system that can process multiple ideas simultaneously, using computational resources to provide optimal solutions. It is the first publicly available multi-agent AI from Google and is accessible to Ultra plan subscribers for $250 per month through the Gemini app.
Gemini 2.5 Deep Think made headlines by securing a gold medal at the International Math Olympiad (IMO) for solving complex mathematical problems. It achieved a score of 34.8% on Humanity’s Last Exam (HLE), outperforming competitors such as xAI’s Grok 4 and OpenAI’s o3. The model integrates with tools like code execution and Google Search, providing detailed and creative responses, which makes it a game-changer for research and development.
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, 2025Google is providing this award-winning model to selected academics and researchers, highlighting its capability to reason for extended periods on complex tasks. The company intends to broaden access through the Gemini API for business and development testing.
Unlike traditional models, Gemini 2.5 Deep Think’s resource-intensive multi-agent approach requires significant infrastructure, funded by premium subscriptions. Its capabilities signal Google’s push to lead in AI innovation, with potential to transform academic and industrial applications.
Gemini 2.5 Deep Think highlights Google’s advancements in artificial intelligence, providing enhanced reasoning capabilities for tackling complex challenges. Its release to testers and academics has the potential to lead to significant breakthroughs across various fields. To stay informed about updates in AI technology, it’s important to rely on trusted sources to avoid misinformation.
