Overview
- Search’s AI mode adds a Thinking option that routes complex questions to Gemini 3 and generates dynamic visual layouts, simulations and tools such as custom calculators.
- Gemini 3 is live across Google Search and the Gemini app, with broader availability to follow, and an experimental Gemini Agent rolls out on the web for AI Ultra subscribers in the U.S.
- Google highlights multimodal reasoning, a million‑token context window and stronger coding performance, enabling tasks like translating handwritten notes, building interactive UIs and long‑horizon planning.
- Developers get preview access via AI Studio, Vertex AI and the new Google Antigravity platform to build autonomous, task‑oriented agents that can coordinate tools like Gmail, Calendar and live web browsing.
- The company cites built‑in safeguards, top community rankings such as LMArena’s No. 1 spot for Gemini 3 Pro, and large user reach for existing features, while publishers warn that AI answers can reduce referral traffic.