Google has launched Gemini 3, the latest version of its large language model (LLM), now available through the Gemini app and Google’s AI-powered search.
According to a blog post published on November 18, Gemini 3 is described as Google’s most capable model to date.
A research-focused version called Gemini 3 Deepthink will also be released soon for Google AI Ultra users, once it finishes extra safety reviews.
Did you know?
Subscribe – We publish new crypto explainer videos every week!
How to Create an NFT: Easiest Way (Animated Explainer)
Tulsee Doshi, Google’s head of product for the Gemini model, said in a blog post:
With Gemini 3, we’re seeing this massive jump in reasoning. It’s responding with a level of depth and nuance that we haven’t seen before.
Early tests show that Gemini 3 scored 37.4 on the Humanity’s Last Exam benchmark. The previous best score, held by GPT-5 Pro, was 31.64. It also led the LMArena benchmark, which measures user satisfaction with AI responses.
The Gemini app currently serves over 650 million users each month, according to Google, and about 13 million developers use its tools in their daily work.
Additionally, Google introduced a new coding tool called Google Antigravity. The system combines features of a chat assistant, command line, and web browser. It allows programmers to write and test code in one place.
Google Antigravity supports a style of “agentic coding”, similar to newer AI-based development tools like Warp and Cursor 2.0.
Google recently introduced its Gemini AI assistant to Maps in India. What features does it offer? Read the full story.
