Google Unveils Gemini 2.5 Deep Think: Advanced Multi-Agent AI Launched

Google has launched Gemini 2.5 Deep Think, its most advanced multi-agent AI model, exclusively for AI Ultra subscribers. Priced at $249.99/month, Deep Think sets a new benchmark in parallel reasoning, outperforming OpenAI and xAI on tests like Humanity’s Last Exam and LiveCodeBench V6. The model’s unique parallel thinking allows it to explore multiple solutions at once, boosting problem-solving power, especially in math and coding challenges. While a research variant won gold at the 2025 International Mathematical Olympiad, the consumer version delivers faster, bronze-level performance on similar tasks.

Read more: TechCrunch

Google DeepMind’s Gemini AI Wins Gold at IMO 2025

In a major AI milestone, Google DeepMind’s Gemini Deep Think model achieved gold medal performance at the International Mathematical Olympiad (IMO) 2025, solving 5 out of 6 problems — scoring 35/42 points. Unlike last year’s models, Gemini worked entirely in natural language, solving problems within the same 4.5-hour limit as human participants. This shows how close AI is getting to elite human reasoning. With advanced reasoning tools like “parallel thinking,” Gemini is now seen as a breakthrough in mathematical AI.

Credit: Deepmind

X