仅凭调用Fable API的日志无法证明Moonshot AI蒸馏了Anthropic的Fable模型用于开发K3。若Anthropic怀疑是蒸馏请求，该请求会被秘密重定向至Opus，Moonshot实际收到的是Opus而非Fable的回复。要证明Fable与K3相关，需提供每次补全的served-model ID、与K3训练任务关联的提取语料库，以及模型血统证据。
There’s also a model-attribution problem here.
Calling the Fable API does not prove Fable generated the response. Even if supposing Moonshot sent a request to the Fable API, if Anthropic suspects that its a distillation request, then the requst will be secretly redirected to Opus.
So Moonshot would then receive an Opus answer, not a Fable answer.
So even if the logs prove Moonshot extracted answers from Claude, but they would not automatically prove those answers came from Fable specifically.
To prove that Fable is related to K3, you’d need to show the served-model ID for each completion, the extracted corpus that is linked to K3 training jobs, and model-lineage evidence like a canary transfer, unique error fingerprints, or an ablation.
We have information that Moonshot AI distilled Anthropic’s Fable for the development of its K3 model. To do this they developed a sophisticated internal platform to conduct large scale distillation against U.S. models, allowing them to quickly switch between multiple methods of access to avoid detection. Moonshot AI has also acquired GB300-equipped servers and has accessed GB300s in Thailand, likely to train its AI models. The United States strongly supports the free and fair development of AI, including a thriving competitive ecosystem that spans frontier models, specialized systems, open-source frameworks, and open-weight models. Legitimate AI distillation used to create smaller, more efficient models plays a vital role in this open innovation ecosystem. However, large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology and undermining American research is unacceptable.在 X 查看被引用的帖子
来源：Rohan Paul · x.com