Cursor releases Composer 2 coding model, benchmarking above Claude Opus 4.6 but below GPT-5.4

Cursor released Composer 2, a proprietary coding model that benchmarks above Claude Opus 4.6 on several tasks but below GPT-5.4, VentureBeat reported. Composer 2 is Cursor's first internally developed model, showing particular strength in multi-file editing and refactoring. The release intensifies competition in AI coding tools, where Cursor competes with GitHub Copilot, Windsurf, and Claude Code.