Google has reignited the AGI race with the launch of Gemini 3, and the tech giant isn’t holding back. The model is being positioned as a major leap forward, with Google claiming benchmark wins over OpenAI’s GPT-5.1 and Anthropic’s Claude Sonnet 4.5 across LMArena, Humanity’s Last Exam, GPQA Diamond, and several advanced reasoning tests. Sundar Pichai calls Gemini 3 “state-of-the-art in reasoning,” and early indicators suggest Google is making its strongest competitive move in years.

The biggest surprise, however, is Google Antigravity, an agent-first development platform built around Gemini 3. Designed to give AI agents full autonomy, it provides access to an editor, terminal, and browser, allowing them to plan, build, and execute software tasks end-to-end. Imagine Cursor, Replit, and autonomous agents fused into a single Google-powered ecosystem.

According to DeepMind leaders Demis Hassabis and Koray Kavukcuoglu, this marks “the beginning of the Gemini 3 era,” with the model now integrated across Search, Workspace, the Gemini app, and developer tools.

With multimodal upgrades, expanded context, improved reasoning modes, and stronger safety performance, Google is positioning Gemini 3 as a decisive step toward AGI. Real-world performance will be the ultimate benchmark, but there’s no doubt the AI leaderboard just shifted.