Google has unveiled Gemini 3 Pro, its most intelligent AI model to date, designed to enhance coding, multimodal understanding, and agentic development. Available through the Gemini API in Google AI Studio and Vertex AI, the model outperforms previous versions in benchmarks and coding tasks, achieving 54.2% on Terminal-Bench 2.0 and topping the WebDev Arena leaderboard with 1487 Elo. Gemini 3 Pro introduces new capabilities such as 'vibe coding', where natural language serves as the only syntax, enabling users to build apps from a single prompt. It also supports advanced multimodal reasoning across text, image, and video, with a 1 million-token context window for complex tasks. The release includes Google Antigravity, a new agentic development platform that lets developers collaborate with autonomous AI agents across editors, terminals, and browsers. Gemini 3 Pro is available in preview for enterprises and free with rate limits in Google AI Studio, marking a major step in AI-driven software creation.