Google announced new AI updates at its I/O conference, including the rollout of its latest Gemini models and a new system designed to simulate the physical world.

The company introduced Gemini 3.5 Flash, a faster and cheaper model that will become the default in its Gemini app and search tools.

It also previewed Gemini Spark, an AI agent that can take actions across connected apps to help users complete tasks with minimal input.

In addition, Google unveiled “Omni,” a world model capable of simulating environments and editing or generating more realistic images and videos, CNBC has reported.

The updates reflect Google’s effort to compete with rivals like OpenAI and Anthropic while expanding more agent-based AI services.