Google has unveiled Gemini 3, its latest suite of advanced AI models, which includes the Gemini 3 Pro, now available across Search, the Gemini app, and various developer tools. This model is described as Google's most intelligent AI yet, featuring superior multimodal understanding and enhanced reasoning.
Performance and Benchmarks
Gemini 3 Pro has achieved top benchmark scores, reaching 1,501 points on the LMArena leaderboard, surpassing Grok 4.1's 1,483 and Gemini 2.5 Pro's 1,452. On Humanity's Last Exam, it scored 37.5% without tool assistance. A higher-variant model, Gemini 3 Deep Think, scores around 41% and is currently undergoing additional safety checks.
Features and Availability
Gemini 3 Pro excels in coding and generative UI functionalities, offering significant benefits in web development and game design. It is available to developers through Google AI Studio, Vertex AI, and third-party platforms such as GitHub and Replit. This AI model also forms the foundation of Google Antigravity, an agentic development environment designed for independent software task execution.
User Access and Rollout
For consumers, Gemini 3 Pro is accessible within the Gemini app and integrated into AI Mode in Google Search for paid users. Wider rollouts with an automatic model selector for Pro and Ultra subscribers are planned.
