Cursor, the AI-powered coding platform, has introduced its first AI-based coding model, Composer, and a multi-agent development environment.
According to the company, Composer is a “frontier” model and works four times faster than other models with similar capabilities. The system, designed for low-latency agent-based coding, can perform most code-related tasks in less than 30 seconds.
Cursor says that the main goal of the model is to work with greater speed and accuracy on large code bases, thereby increasing programmer productivity. To do this, Composer has been trained with a wide range of tools — among which the “semantic search across the code base” function stands out. This function helps the model better understand complex projects.
The new version of the platform has moved from the traditional “file-centric” structure to an “agent-centric” environment. This approach allows programmers to focus on working with the result, not on the details of the code.
At the same time, the classic IDE interface and file viewing mode have been retained. Thanks to the simplified interface, users can more easily track changes made by AI agents.
A built-in browser tool added to the platform allows AI agents to automatically test and make corrections to the code they write. According to Cursor, this feature shows that AI is becoming a system that not only writes code, but also checks and improves its own work.
