Google has introduced the Gemini 3 AI model, an update of Gemini with improved visual reasoning, and Google Antigravity, an agentic development platform for AI-assisted software development.
Both were announced on November 18. Gemini 3 is positioned as offering reasoning capabilities with robust function calling and instruction adherence to build sophisticated agents. Agentic capabilities in Gemini 3 Pro are integrated into agent experiences in tools such as Google AI Studio, Gemini CLI, Android Studio, and third-party tools. Reasoning and multimodal generation enable developers to go from concept to a working app, making Gemini 3 Pro suitable for developers at any experience level, Google said. Gemini 3 surpasses Gemini 2.5 Pro at coding, mastering both agentic workflows and complex zero-shot tasks, according to the company. Gemini 3 is available in preview at $2 per million input tokens and $12 per million output tokens for prompts of 200K tokens or less through the Gemini API in Google AI Studio and Vertex AI for enterprises.
Google Antigravity, meanwhile, shifts from agents embedded within tools to an AI agent as the primary interface. It manages surfaces such as the browser, editor, and the terminal to complete development tasks. While the core is a familiar AI-powered IDE experience, Antigravity is evolving the IDE toward an agent-first future with browser control capabilities, asynchronous interaction patterns, and an agent-first product form factor, according to Google. Together, these enable agents to autonomously plan and execute complex, end-to-end software tasks. Available in public preview, Antigravity is compatible with Windows, Linux, and macOS.
