Particle.news

Google Launches Gemini 3 With Day-One Search Integration and Antigravity Dev Platform

Google puts the new model into Search on day one to deliver a more visual, agent-ready experience.

Overview

  • Gemini 3 Pro is now live in the Gemini app and available to developers via AI Studio, Vertex AI, APIs, the Gemini CLI and the new Antigravity environment, with paid access in Search’s AI Mode beginning in initial markets.
  • Search’s AI Mode now uses Gemini 3 to generate immersive layouts with interactive tools and simulations, aiming to reduce follow-up prompts by better inferring user intent.
  • The Deep Think variant focused on advanced reasoning remains under security review and is limited to testing by security researchers and Google AI Ultra users for now.
  • Google unveiled Antigravity, an agentic IDE where AI agents can plan and execute coding tasks with access to an editor, terminal and browser, integrating Computer Use and image generation models.
  • Google and DeepMind cite top-tier benchmark results such as LMArena and emphasize safety improvements like greater resistance to prompt injection, as the company reports 650 million monthly users of the Gemini app and early coverage notes strong market reception.