Cursor's Composer 2.5 matches Opus 4.7 and GPT-5.5 benchmarks at a fraction of the cost

Cursor released Composer 2.5, a new in-house AI coding model built on Moonshot’s open-source Kimi K2.5 checkpoint.

The model was trained with more synthetic tasks and reinforcement learning, and Cursor says it matches Opus 4.7 and GPT-5.5 on SWE-Bench Multilingual and CursorBench v3.1.

Cursor claims Composer 2.5 delivers frontier-level coding performance at much lower per-token cost.

The company also said it is training a much larger successor model from scratch with SpaceX and xAI on the Colossus-2 cluster.