Cursor has introduced Composer 2, a new artificial intelligence model for programming that is already built into the company’s development environment. The company says the system can handle complex coding work and complete long chains of actions more effectively than earlier versions. Reports Devby.io.
According to Cursor, the new model shows a clear jump in quality over previous releases. In the company’s internal CursorBench test, Composer 2 scored 61.3 points, compared with 44.2 for Composer 1.5 and 38.0 for the first version.
Cursor also reported gains in Terminal Bench 2.0 and SWE-bench Multilingual, two benchmarks used to measure how well AI agents operate in terminal-based workflows and solve software tasks across different programming languages.
The company is putting special emphasis on price-to-performance. The base version of Composer 2 is priced at about $0.50 per million input tokens and $2.50 per million output tokens. A faster version is also available at roughly $1.50 and $7.50, while Cursor says it still remains cheaper than many rival tools for developers.
Composer 2 is now part of Cursor’s ecosystem and is available through individual subscription plans, where it uses a separate resource pool with an included quota. The model is also undergoing early testing in the company’s new user environment.
Read “Zamin” on Telegram!
