Cursor admits its new coding model uses Moonshot AI's Kimi

AI coding company Cursor launched its new Composer 2 model, promoting it as offering high-level coding intelligence. However, the company faced immediate scrutiny from an X user posting under the name Fynn, who alleged that Composer 2 was simply Kimi 2.5 with additional reinforcement learning—an AI training method based on trial and error. Kimi 2.5 is an open-source model recently released by Moonshot AI, a Chinese company backed by Alibaba and HongShan. As evidence, Fynn pointed to code that appeared to identify Kimi as the model, scoffing that the company should have at least renamed the model identifier.

The revelation raised questions because Cursor is a U.S. startup. The company raised a $2.3 billion funding round last fall at a $29.3 billion valuation, and is reportedly exceeding $2 billion in annualized revenue. Despite these resources, the company did not mention Moonshot AI or Kimi in its initial announcement, leading to questions about its reliance on a Chinese model. Cursor’s vice president of developer education, Lee Robinson, acknowledged that Composer 2 started from an open-source base. However, Robinson defended the model’s performance, asserting that only ~1/4 of the compute spent on the final model came from the base, with the rest coming from Cursor’s own training.

Robinson also asserted that Cursor’s use of Kimi was consistent with the terms of its license. This point was supported by the Kimi account on X, which posted that Cursor used Kimi as part of an authorized commercial partnership with Fireworks AI. The Kimi account welcomed the integration of its model through Cursor’s continued pretraining—the initial phase of training an AI model on a large dataset—and reinforcement learning. Still, Cursor co-founder Aman Sanger admitted the lack of upfront disclosure was a mistake. Sanger stated, “It was a miss to not mention the Kimi base in our blog from the start. We’ll fix that for the next model.”

Building on top of a Chinese model carries geopolitical weight amid the ongoing AI technology competition, which is often framed as an existential battle between the United States and China. This tension was recently highlighted when Silicon Valley experienced concern after the Chinese company DeepSeek released a competitive model early last year.

Why it matters

The reliance of a high-profile U.S. startup on a Chinese AI model highlights the complexities of the global AI supply chain and the growing pressure for transparency in model provenance.