Google has officially launched Gemini 3, its most advanced and capable foundation model to date. The new artificial intelligence (AI) is immediately accessible through the Gemini app and Google's AI search interface, marking a significant leap in AI capabilities.

The release of Gemini 3 comes a mere seven months after its predecessor, Gemini 2.5, highlighting the rapid pace of innovation in the frontier AI space. This latest iteration is positioned as Google's most powerful large language model (LLM) yet, immediately contending for the title of the most capable AI tool available on the market. Its debut also follows closely on the heels of OpenAI's GPT 5.1 and Anthropic's Sonnet 4.5, underscoring the intense competition and accelerated development in the AI sector.

"With Gemini 3, we're seeing this massive jump in reasoning," stated Tulsee Doshi, Google's head of product for the Gemini model. "It's responding with a level of depth and nuance that we haven't seen before."

Gemini 3's enhanced reasoning power is already reflected in independent evaluations. The model achieved a record-breaking score of 37.4 on the Humanity's Last Exam benchmark, which is designed to assess general reasoning and expertise. This surpasses the previous high score of 31.64 held by GPT-5 Pro. Furthermore, Gemini 3 topped the LMArena leaderboard, a human-led benchmark that measures user satisfaction, indicating strong real-world performance.

Google reports widespread adoption of its AI technology, with the Gemini app currently boasting over 650 million monthly active users. Additionally, 13 million software developers have integrated the Gemini model into their workflows, leveraging its capabilities for various programming tasks.

Introducing Google Antigravity: AI-Powered Coding

Alongside the core Gemini 3 model, Google also unveiled Google Antigravity, a new Gemini-powered coding interface. This innovative tool facilitates multi-pane agentic coding, drawing comparisons to advanced agentic integrated development environments (IDEs) such as Warp or Cursor 2.0.

Antigravity uniquely combines a ChatGPT-style prompt window with a command-line interface and a browser window. This setup allows developers to observe the immediate impact of changes suggested or implemented by the AI coding agent, streamlining the development process.

"The agent can work with your editor, across your terminal, across your browser to make sure that it helps you build that application in the best way possible," explained Koray Kavukcuoglu, DeepMind CTO.

Looking ahead, a more research-intensive variant of the model, dubbed Gemini 3 Deepthink, is slated for release to Google AI Ultra subscribers in the coming weeks, pending further rounds of safety testing.