The new model is now available to both regular users and subscribers, and according to the company, it surpasses the previous 2.5 Pro version across all major benchmarks. Gemini 3 Pro is designed to handle complex tasks more accurately and, notably, focuses on providing what the user truly needs, rather than simply what they want to hear.
Through the “AI Mode” in Google Search, the model better understands the nuances of queries and offers more interactive responses. In the Gemini app, answers have become more concise and useful, while the functionality of apps created through the “Canvas” tool has been enhanced. Users can access these features by selecting the “Thinking” mode from the menu, and subscribers are offered higher usage limits.
The company has also launched the “Gemini 3 Deep Think” model, which features strengthened reasoning abilities. It has demonstrated strong results in challenging evaluations across biology, physics, and general intelligence. Another major update is the “Antigravity” platform—an integrated environment for writing and testing code using AI agents. The platform is currently available for free in a trial mode.
