In a move that reshapes the competitive landscape among AI giants, Google announced on Tuesday the launch of its groundbreaking Gemini 3 model — described as the most powerful model the company has ever built. The release comes at a heated moment in the industry, just days after OpenAI’s GPT-5.1 and two months following Anthropic’s Sonnet 4.5, further fueling the race among advanced AI systems.
Google Unveils Gemini 3: A Massive Leap in Reasoning and Depth
Tulsi Doshi, Head of Product for Gemini at Google, stated that the new model delivers an unprecedented leap in reasoning and precision, offering a new level of analytical depth not seen in previous versions.
Gemini 3 arrives only seven months after Gemini 2.5 and is now available immediately through the Gemini app and Google's AI-enhanced search interface.
Google is also preparing to launch a deeper research-focused version called Gemini 3 Deepthink, expected to roll out to "Google AI Ultra" subscribers after additional safety reviews are completed.
Record-Breaking Performance: Gemini 3 Surpasses Industry Leaders
The new model delivered impressive results across major global AI benchmarks:
-
37.4 points on Humanity’s Last Exam — the highest score ever recorded, beating GPT-5 Pro’s 31.64.
-
Ranked first on LMArena, the human evaluation platform measuring user satisfaction.
Google also revealed that the Gemini app has exceeded 650 million monthly active users, and more than 13 million developers rely on the model in their work.
Google Antigravity: A New Coding Platform Powered by AI Agents
Alongside the model launch, Google introduced a fully integrated coding environment named Google Antigravity, built on Gemini 3 and designed to deliver a new multi-window development experience similar to tools like Cursor 2.0 and Warp.
