Google Launches Gemini 3 AI Model With Advanced Features -

Google unveiled Gemini 3 on November 18, 2025, its most intelligent AI model yet, available across the Gemini app, Google Search’s new AI Mode, the Gemini API via AI Studio/Vertex AI, and enterprise subscriptions. The Gemini 3 family includes a 'Pro' version for general use and a 'Deep Think' variant (hinted for complex reasoning). The Pro model has achieved record-breaking scores on major benchmarks, surpassing GPT-5 and Claude Sonnet 4.5. Google claims significant improvements in mathematics, coding, spatial reasoning, and multimodal capabilities—processing text, vision, and tools simultaneously. Integration from day one rivals previous Google launches, signaling a strategic shift to deploy top-tier LLMs at scale.

For developers, Gemini 3 introduces Antigravity, a coding tool that lets AI agents access a code editor, terminal, and browser together for multi-step workflows. The model also excels in tool usage and long-term planning. This launch comes amid fierce competition: OpenAI recently released GPT-5.1, and xAI unveiled Grok 4.1. Alphabet recently increased its AI capital expenditure budget from $91 billion to $93 billion. Gemini 3 will first roll out to Pro and Ultra subscribers in the US before expanding globally. Its success hinges on adoption by developers, enterprises, and end-users, not just benchmark leadership.