Google Launches Gemini 3 as AI Race with OpenAI Heats Up

Google has introduced Gemini 3 , its newest artificial intelligence model, stepping up its rivalry with OpenAI’s GPT-5. Alphabet CEO Sundar Pichai said the model is designed to answer more complex questions with fewer prompts, better understanding context and intent, so people “get what they need with less prompting.” Gemini 3 will appear in the Gemini app, Google’s AI search features AI Mode and AI Overviews, and in business tools, with access starting Tuesday for select paid users and expanding soon.

The launch comes about eight months after Gemini 2.5 and less than a year after Gemini 2.0, showing how quickly Google is updating its AI stack. Pichai said AI has gone in two years from reading text and images to “reading the room,” and that the company is now delivering Gemini at full Google scale. The Gemini app has 650 million monthly users, and AI Overviews has 2 billion, while OpenAI says ChatGPT has 700 million weekly users.

Google says Gemini 3 aims to move away from flattery and generic answers toward clearer, more honest responses, responding to critics who say chatbots are too eager to please. Demis Hassabis, who leads Google DeepMind, said the system is meant to trade cliché for “genuine insight.”

The company also revealed Google Antigravity, an “agent” platform that lets developers work at a higher, task-focused level instead of writing every line of code. Executives describe Gemini 3 as Google’s best “vibe coding” model so far, referring to tools that turn plain-language prompts into software.

Gemini 3 will power new “generative interfaces” that can present answers like a digital magazine, mixing text, images, tables, and interactive elements. In AI Mode, subscribers could ask for an explanation of an art gallery or a physics idea and get a rich layout, an interactive loan calculator, or a small simulation.

Through the Gemini API and Google Cloud’s Vertex AI, businesses will be able to plug Gemini 3 into their own systems. Google says corporate users can use it to build onboarding and training materials, inspect videos and factory images more accurately, and manage tasks like procurement as tech giants pour more than $380 billion this year into AI infrastructure. The launch shows how hard Google and OpenAI now compete.