Google has unveiled Gemini 3, the latest upgrade to its large language model and a major step in its efforts to close the gap with OpenAI. The new model promises more sophisticated answers in the Gemini app, improved performance in AI Mode on Google Search and stronger capabilities in coding, app development and image generation.

Koray Kavukcuoglu, chief technology officer at Google DeepMind, said the company hopes the model will help users turn any idea into reality. Anticipation for the launch had been building for weeks, with online communities on Reddit and X hunting for signs that Google had quietly rolled out early features.

Gemini’s popularity has surged this year, driven in large part by Nano Banana, Google’s fast new image generator. The tool has created more than five billion i

See Full Page