Google has unveiled a new AI rival, Gemini 3, hailed as its “most advanced model.” This extensive language model (LLM) features improved reasoning abilities, enabling it to comprehend tasks’ intricacies and subtleties more effectively. It is crafted to understand the context and intent behind requests, minimizing the necessity for multiple prompts. Gemini 3 has been integrated into Google’s offerings, such as Google Search, AI Studio, Vertex AI, and the Antigravity platform. Originally accessible to Google AI Pro and Ultra subscribers, it signifies the first instance of a new model being launched in Search on the same day. Gemini 3 Pro tops the LMArena with an Elo score of 1501, outpacing xAI’s Grok and its forerunner, Gemini 2.5 Pro. For more intricate tasks, Gemini 3 Deep Think provides enhanced reasoning capabilities. Google has also redesigned the Gemini app, incorporating features such as the My Stuff folder for better chat and document organization, improving the shopping experience, and testing generative interfaces. The Gemini Agent function can manage multi-step tasks like scheduling appointments, organizing emails, and performing online research.