Google unveils Gemini 3, bringing advanced AI to Search and the Gemini app
Google is taking the fight to AI rival ChatGPT with the launch of Gemini 3, the search giant's new flagship AI model built to handle complex reasoning, interpret multiple types of media and power new agent features across Search, the Gemini app, developer tools and enterprise platforms.
In a note announcing the new model, Google and Alphabet CEO Sundar Pichai said, “Nearly two years ago we kicked off the Gemini era, one of our biggest scientific and product endeavors ever undertaken as a company. Since then, it’s been incredible to see how much people love it.”
He highlighted scale across the ecosystem, noting that AI Overviews serve “2 billion users every month” and the Gemini app “surpasses 650 million users per month.” In comparison, OpenAI's CEO Sam Altman claimed in October that ChatGPT boasts 800 million weekly users.
Pichai added, “more than 70 percent of our Cloud customers use our AI” and “13 million developers have built with our generative models.”
Pichai described how the series has progressed. “Gemini 1’s breakthroughs in native multimodality and long context window expanded the kinds of information that could be processed -- and how much of it.” He added that “Gemini 2 laid the foundation for agentic capabilities and pushed the frontiers on reasoning and thinking” and pointed to “Gemini 2.5 Pro topping LMArena for over six months.”
Gemini 3 is viewed as a significant step up. “We’re introducing Gemini 3, our most intelligent model, that combines all of Gemini’s capabilities together so you can bring any idea to life,” Pichai said, adding that it has “state of the art in reasoning” and is “built to grasp depth and nuance.”
It is reportedly better at understanding user intent. “It’s amazing to think that in just two years, AI has evolved from simply reading text and images to reading the room,” Pichai said.
Gemini 3 is available immediately in AI Mode in Search, in the Gemini app, in AI Studio, in Vertex AI and in the new agentic development platform Antigravity.
Gemini 3 Pro
The detailed technical rollout begins with Gemini 3 Pro in preview. Google claims it can outperform 2.5 Pro on leading benchmarks in reasoning, mathematics and multimodal understanding, as well as interpret long documents, break down research material, extract structure from media and generate interactive learning tools.
A second mode, Gemini 3 Deep Think, extends the model's reasoning further and can apparently post higher scores on tests designed to evaluate novel problem solving. This will be released to Google AI Ultra subscribers after "additional safety evaluation".
Google is also launching Antigravity, a new agent driven development environment that allows agents to work directly in an editor, terminal and browser to plan and complete coding tasks across entire projects.
In consumer products, AI Mode in Search is now able to produce dynamic layouts and interactive visual explanations using Gemini 3. Not everyone is fan of AI being jammed into Google Search, so it will be interesting to see if the new model can win them over. The Gemini app's agent abilities have been expanded to carry out multi step workflows such as inbox organization and service booking.
Google says Gemini 3 has undergone its most extensive safety evaluation to date, and more models in the Gemini 3 series are set to arrive in the coming weeks.
What do you think about Google's new AI model? Let us know in the comments.