Google released Gemini 3 today, and the company is pitching it as a major leap forward in its AI lineup. The new model features better reasoning, stronger multistep agent abilities, and improved “vibe coding,” which is Google’s way of saying the responses feel cleaner and more intentional.

Google says users should see answers that are more concise and better formatted. The interface also received an update, and the shopping experience has been described as “radically” improved.

The company introduced a feature called Generative Interfaces. It can turn a response into a magazine-style layout or build an interactive presentation from a simple query. Both options are still labeled as experiments.

Another experimental feature is Gemini Agent. It is designed to handle complex, multistep tasks across Google apps without leaving the Gemini environment.

Google wants to use its ecosystem to turn Gemini into the default AI assistant. Early benchmark results help the case. The company says Gemini 3 topped GPT-5.1 and Claude Sonnet 4.5 on many popular tests and jumped into the number one spot on the LMArena leaderboard. $GOOGL ( ▼ 1.15% )