Overview
- Gemini 3 Pro is available now in the Gemini app for all users, with access in Google Search’s AI Mode for paying AI Pro and Ultra subscribers and broader rollout promised in the coming weeks.
- The model emphasizes stronger reasoning and native multimodality, generating dynamic interfaces such as magazine-style layouts, interactive tools, and real-time simulations in responses.
- An experimental Gemini Agent can perform multi-step tasks and act in Gmail, Calendar, and Reminders with user approval, launching first for AI Ultra subscribers inside the app.
- Developers can use Gemini 3 today via Google AI Studio, Vertex AI, Gemini CLI, and the Gemini API, alongside Antigravity, a new agentic platform where AI agents plan, execute, and validate coding tasks.
- Google cites top benchmark results, including a 1,501 score on LMArena and a lead on WebDev Arena, while a higher-reasoning Deep Think mode enters safety testing ahead of release to subscribers.