Overview
- Gemini 3 Pro is available today in the Gemini app and in Search’s AI Mode via the “Thinking” picker for AI Pro and Ultra subscribers in the US, with broader AI Mode access and automatic routing to Gemini 3 planned in the coming weeks.
- The model introduces generative interfaces that create dynamic layouts and interactive tools, including a magazine-style visual layout and a Dynamic View that codes custom UIs for queries such as travel plans or learning tasks.
- An experimental Gemini Agent, powered by Gemini 3, can handle multi‑step tasks and interact with Google apps like Calendar and Gmail, launching first for AI Ultra subscribers inside the Gemini app.
- Google unveiled Antigravity, an agentic IDE where AI agents operate across editor, terminal, and browser to plan and execute software tasks; it is available in preview on Mac, Windows, and Linux.
- Google cites top benchmark results for Gemini 3 Pro (e.g., 1501 on LMArena, 37.5% on Humanity’s Last Exam without tools, 91.9% on GPQA Diamond) and says the higher‑scoring Deep Think mode is limited to safety testers before reaching AI Ultra subscribers.