Cursor has released a major update — Cursor 2.0 and its first coding model called Composer.
Composer is a frontier model specifically designed for agentic coding. It is 4 times faster than its counterparts and can complete complex multi-step tasks in less than 30 seconds. The model is trained using codebase-wide semantic search, making it particularly effective when working with large codebases.
Cursor 2.0 also introduces a new multi-agent interface. Now, work is organized around agents rather than files: you can run multiple agents in parallel using git worktrees or remote machines. An interesting effect is that the simultaneous work of several models on one task improves the quality of the final result.
The company has also added tools for quick code review and an integrated browser for auto-testing changes. It is noticeable that Cursor is becoming not just an IDE, but a full-fledged environment for multi-model work with AI agents.
🔗 Source: https://cursor.com/blog/2-0
🔗 Download: https://cursor.com/download
