Google has updated its Gemini 3 Deep Think reasoning model to improve its performance on complex problems where data is chaotic or incomplete, performing at the level of a gold medalist at the International Mathematical Olympiad.
Google has updated its Gemini 3 Deep Think reasoning model to improve its performance on complex problems where data is chaotic or incomplete, performing at the level of a gold medalist at the International Mathematical Olympiad.
Google says that Deep Think combines deep scientific knowledge with «engineering utility,» giving the model more options for practical application.
«Last year, we demonstrated that specialized versions of Deep Think can successfully solve the most complex reasoning tasks, achieving gold standards at the World Math and Programming Championships. Recently, Deep Think has enabled specialized agents to conduct research in mathematics at a scientific level,» the developers said.
Gemini 3 Deep Think generated a real-time 3D WiFi radar that maps every network around you as glowing nodes in a Matrix-style space — in one shot. It used Pearson correlation to infer which APs are physically close, since RSSI alone isn’t enough. pic.twitter.com/xJ3Jc8tr6j
— BijanBowen (@Ominousind) February 13, 2026
Google highlights the following achievements of Gemini 3 Deep Think in tests:
The new Deep Think is now available in the Gemini app for Google AI Ultra subscribers. Google is also making Deep Think available through the Gemini API for select researchers, engineers, and enterprises.
