Cursor Launches Composer 2: Faster, Cheaper, and Better at Complex Coding

What happened

Cursor, the AI coding platform valued at $29.3 billion, has launched Composer 2, its new in-house AI model built specifically for software development. The model is now live inside Cursor and delivers significantly better performance than its predecessor, Composer 1.5, at a fraction of the cost.

Composer 2 Standard is priced at $0.50 per million input tokens, down from $3.50. That is an 86% price cut. A faster variant, Composer 2 Fast, is also now the default experience for Cursor users. On benchmarks, Composer 2 outperforms Anthropic's Claude Opus 4.6, though it still trails OpenAI's GPT-5.4 at the top end.

The key technical claim is that Composer 2 is better suited to long-horizon coding, meaning it can handle complex tasks that require hundreds of steps, not just short code completions. Cursor trained the model on long-horizon coding tasks specifically to improve this capability, and the results represent a 50% reduction in so-called "compaction errors," where AI models forget context mid-task.

What this means for your business

Cursor is already one of the most popular AI coding tools among professional developers. If your business relies on software development, whether in-house or through an agency, Composer 2 makes Cursor more capable and more cost-effective than before.

The broader story here is that the AI coding tools market is intensely competitive right now, and prices are falling while performance improves. If you have not reviewed which AI tools your developers are using recently, now is a good time. The right tools can meaningfully speed up development cycles and reduce costs.