Google Launches Gemini 3 With Day-One Search Integration and New AI Agents

Overview

• Availability begins today for users and developers, with the new model embedded in Google Search from launch.
• Paying subscribers gain Gemini 3 capabilities in the Search AI Mode, and the Gemini app now returns interactive, site‑like responses.
• Google describes Gemini 3 as a multimodal model spanning text, images, audio and code, promoted as its most capable system to date.
• The company touts record benchmark scores and “PhD‑level” reasoning on tests it cites, including Humanity’s Last Exam and GPQA Diamond.
• New tools include Gemini Agent for multi‑step tasks such as inbox triage and trip booking, Gemini 3 Pro for more complex work, and Antigravity for autonomous software development workflows in enterprises.