Google DeepMind Introduces Gemini 3, AI Model That Is More Advanced And Smart

Gemini 3 (photo: Google)

JAKARTA - Not long ago releasing Gemini 2.5 Pro, Google DeepMind now introduces Gemini 3, Gemini's most intelligent model to help you come up with various ideas.

Gemini 3 is designed to bring the most advanced reasoning skills to understand depth and feel when capturing small details in a creative idea or unraveling the complex layer of complex problems.

Not only that, the search giant also stated that Gemini 3 is much better at understanding the context and meaning behind your request, so you can get the right answer with less prompt.

"Today we are taking a big step on the way to Artificial General Intelligence (AGI) by releasing Gemini 3," said Google DeepMind CEO, Demis Hassabis, and Google DeepMind CTO and Chief AI Architect, Google, Koray Kavakcuoglu, in its official blog.

They claim that Gemini 3 is the best AI model in the world for a multimodal understanding, as well as their most advanced agent model and vibe coding.

"Gemini 3 Pro is able to realize various ideas thanks to its most advanced reasoning and multimodal capabilities. This model significantly outperforms Gemini 2.5 Pro in all major AI benchmarks," he added.

Google also introduced Gemini 3 Deep Think, a more sophisticated mode of reasoning to boost Gemini 3's performance while providing early access for security testers before finally becoming available to Google AI Ultra subscribers.

Starting today, Gemini 3 is already present at AI Mode in the search, on the Gemini app, at AI Studio and Vertex AI for developers, and on Google's latest agency development platform, Google Antigravity.