Google is launching Gemini 3, a new series of models that the company calls its “smartest” and “actually most accurate” AI system. The release is a chance for Google to jump ahead of competitors, including OpenAI, after the troubled launch of GPT-5, and potentially put the company at the forefront of consumer AI models, The Verge reports.
For the first time, Google is making its new flagship model — Gemini 3 Pro — publicly available on the Gemini app on day one. The model is also becoming available to subscribers on the search engine.
Tulsi Doshi, senior director and head of product at Google DeepMind, said the new model will bring the company closer to making information “universally accessible and useful” as search evolves.
The Gemini 3 Pro is “natively multimodal,” meaning it can process text, images, and audio all at once, rather than separately. For example, Google says the Gemini 3 Pro can translate photos of recipes into a full-fledged cookbook or create interactive learning cards based on a series of video lectures.