Google has unveiled Gemini 3, its latest and most sophisticated artificial intelligence model, and is now beginning an extensive deployment across its product ecosystem.
The model is currently accessible to public users in the Gemini app and to Google AI Pro and Ultra subscribers utilising AI Mode in Search. Developers can access the Gemini API via AI Studio, the new Google Antigravity agentic development platform, and Gemini CLI. Enterprise clients can access the model via Vertex AI and Gemini Enterprise.
Google characterises Gemini 3 as its most proficient model to date, exhibiting enhanced reasoning capabilities and a superior understanding of complex requests. The company asserts that it surpasses the prior Gemini 2.5 Pro edition in common standards for coding, mathematics, and multimodal jobs.
In addition to the primary model, Google is developing an advanced iteration, Gemini 3 Deep Think, which will undergo further safety assessments before being released to Google AI Ultra members. The business also highlighted that Gemini 3 had passed extensive safety evaluations.
