Cursor AI's new Composer 2 model has been introduced

The Cursor platform, well-known in the programming field, has introduced its new artificial intelligence model. Zamin.uz reported on this news.

This update is known under the name Composer 2 and has already been integrated into the company's main software environment. Developers emphasize that the new system demonstrates significantly higher efficiency compared to previous versions when writing complex code.

Particular emphasis is placed on its ability to complete time-consuming programming processes faster and without errors. According to the results of internal tests conducted by the company, the new model has achieved significant qualitative growth.

In the CursorBench test, the Composer 2 model scored 61.3 points, while the previous version 1.5 scored 44.2 points and the first version only achieved 38.0 points. At the same time, improvements have also been recorded in international standard tests such as Terminal Bench 2.0 and SWE-bench Multilingual.

This benchmark measures the efficiency of AI agents working in a team and their level of problem-solving across various programming languages. The new model also holds an advantage over competitors in financial terms.

The base version is priced at half a dollar per million input tokens, while output tokens are valued at $2.50. Faster variants are offered at prices of one and a half dollars and seven and a half dollars, respectively.

Company representatives emphasize that these prices represent a serious advantage for developers compared to other competitive tools and in terms of the productivity ratio. Currently, Composer 2 has become an integral part of the Cursor ecosystem and is offered to users through separate subscription plans.

Thanks to using specially allocated resources and having an independent quota, the operation process continues uninterrupted. The model is currently undergoing testing with a new user interface and will serve to further simplify the programming process in the future.