Google has officially launched the new version of its smart models under the name Gemini 3, replacing the Gemini 2.5 that debuted in May during the Google I/O event.

The new model will begin arriving via AI Mode within Google Search for Google AI Pro and AI Ultra subscribers, and is also available in the Gemini app for everyone. Executive Director Sundar Pichai emphasizes that Gemini 3 is becoming more able to understand context and user intent, reducing the need for repeating directions.

Google describes the Gemini 3 as the world's best model for understanding multimedia, offering deeper interaction and richer visual representations. The launch begins with the Gemini 3 Pro version, which is now available with a preview version, followed by the Gemini 3 Deep Think version, which is currently undergoing safety testing before being later made available to Google AI Ultra subscribers.