<p id="speakable-summary" class="wp-block-paragraph">AI coding company Cursor launched a new model this week called Composer 2, which it <a href="https://cursor.com/blog/composer-2" target="_blank" rel="noreferrer noopener nofollow">promoted</a> as offering “frontier-level coding intelligence.” </p>
<p class="wp-block-paragraph">However, an X user posting under the name Fynn <a href="https://x.com/fynnso/status/2034706304875602030" target="_blank" rel="noreferrer noopener nofollow">soon claimed</a> that Composer 2 was “just Kimi 2.5” with additional reinforcement learning — Kimi 2.5 being <a href="https://techcrunch.com/2026/01/27/chinas-moonshot-releases-a-new-open-source-model-kimi-k2-5-and-a-coding-agent/">an open source model recently released by Moonshot AI</a>, a Chinese company backed by Alibaba and HongShan (formerly Sequoia China). </p>
<p class="wp-block-paragraph">As evidence, Fynn pointed to code that seemed to identify Kimi as the model.</p>
<p class="wp-block-paragraph">“[A]t least rename the model ID,” they scoffed.</p>
<p class="wp-block-paragraph">It was a surprising revelation, since Cursor is a well-funded U.S. startup that <a href="https://techcrunch.com/2025/11/13/coding-assistant-cursor-raises-2-3b-5-months-after-its-previous-round/">raised a $2.3 billion round last fall</a> at a $29.3 billion valuation, and is <a href="https://techcrunch.com/2026/03/02/cursor-has-reportedly-surpassed-2b-in-annualized-revenue/">reportedly exceeding $2 billion in annualized revenue</a>. Also, the company didn’t mention anything about Moonshot AI or Kimi in its announcement.</p>
<p class="wp-block-paragraph">However, Cursor’s vice president of developer education <a href="https://x.com/leerob/status/2035035355364081694?s=46" target="_blank" rel="noreferrer noopener nofollow">Lee Robinson soon acknowledged</a>, “Yep, Composer 2 started from an open-source base!” But he said, “Only ~1/4 of the compute spent on the final model came from the base, the rest is from our training.” As a result, he said Composer 2’s performance on various benchmarks is “very different” from Kimi’s.</p>
<p class="wp-block-paragraph">Robinson also insisted that Cursor’s use of Kimi was consistent with the terms of its license, a point the Kimi account on X repeated in a <a href="https://x.com/Kimi_Moonshot/status/2035074972943831491" target="_blank" rel="noreferrer noopener nofollow">subsequent post congratulating Cursor</a>, where it said Cursor used Kimi “as part of an authorized commercial partnership” with Fireworks AI.</p>
<p class="wp-block-paragraph">“We are proud to see Kimi-k2.5 provide the foundation,” the Kimi account said. “Seeing our model integrated effectively through Cursor’s continued pretraining & high-compute RL training is the open model ecosystem we love to support.”</p>
<p class="wp-block-paragraph">So why not acknowledge Kimi upfront? Beyond any potential embarrassment in not creating a model from scratch, building on top of a Chinese model might feel particularly fraught right now, with the so-called AI “arms race” often framed as <a href="https://techcrunch.com/2025/07/23/trumps-ai-strategy-trades-guardrails-for-growth-in-race-against-china/">an existential battle between the United States and China</a>. (See, for example, Silicon Valley’s <a href="https://techcrunch.com/2025/01/30/how-deepseek-changed-silicon-valleys-ai-landscape/">apparent panic after Chinese company DeepSeek released a competitive model</a> early last year.)</p>
<p class="wp-block-paragraph">Cursor co-founder Aman Sanger <a href="https://x.com/amanrsanger/status/2035079293257359663" target="_blank" rel="noreferrer noopener nofollow">acknowledged</a>, “It was a miss to not mention the Kimi base in our blog from the start. We’ll fix that for the next model.”</p>
© 2026 Winzheng.com 赢政天下 | 转载请注明来源并附原文链接