Cursor has announced a major update to its AI coding agents, enabling them to test their own changes, record work via videos and logs, and run in parallel on cloud-based virtual machines. This allows developers to run 10–20 agent tasks simultaneously with high throughput. About 35% of Cursor's pull requests are now generated by these autonomous agents. The update comes as competition intensifies: Anthropic's Claude Code has reached $2.5B in run-rate revenue, OpenAI's Codex surpassed 1.5M weekly active users, and GitHub Copilot claims 26M users. Cursor, valued at $29.3B with over $1B in annualized revenue, positions this as a fundamental shift in how developers work with agents rather than just a new feature.
Sort: