Google announces Gemini 3 model alongside major AI adoption milestones

Google describes Gemini 3 as its most intelligent model designed to bring ideas to life. [1] Every generation of Gemini has built on the previous iterations to expand user capabilities. [1] Gemini 1 introduced breakthroughs in native multimodality and long context windows, which expanded the type and volume of information that the system could process. [1] The subsequent Gemini 2 model laid the foundation for agentic capabilities while pushing the frontiers of reasoning and thinking. [1]

Alphabet CEO Sundar Pichai stated that AI Overviews currently have 2 billion users every month. [1] The Gemini application has surpassed 650 million monthly users, and more than 70 percent of Google Cloud customers currently utilize the company's AI. [1] Additionally, 13 million developers have built applications using Google's generative models. [1] Pichai attributed the company's deployment speed to a differentiated full stack approach encompassing infrastructure, research, tooling, and consumer products. [1]

The announcement of Gemini 3 marks a continuation of Google's rapid iteration cycle in the generative AI space, building sequentially on the multimodal and agentic foundations of Gemini 1 and 2. By highlighting the massive scale of its user base—such as the 2 billion monthly users for AI Overviews and 650 million for the Gemini app—Google is emphasizing its significant distribution advantage in consumer AI products. This massive footprint contrasts with standalone AI platforms, anchoring Google's competitive strategy in its existing full-stack infrastructure and broad cloud adoption. What the sources don't address: when Gemini 3 will be officially released to the public or what specific architectural changes and benchmark improvements the new model features over Gemini 2.