Cursor recently launched its new model, Composer 2, claiming it delivers “frontier-level coding intelligence.” However, users quickly noticed similarities to Kimi 2.5, an open-source model developed by Moonshot AI. This sparked debate after evidence suggested the model may have been built on top of Kimi without initial disclosure.

Cursor later confirmed that Composer 2 did indeed start from an open-source base, but emphasized that most of the final model’s performance came from additional training and reinforcement learning. The situation highlights growing scrutiny around how AI companies build and market models, especially in a competitive global landscape.