Kimi’s cross–data center decoupling this round has something real—only once inference costs are driven down does it truly become for everyone.

ME News message, April 18 (UTC+8), the Moonshot AI team recently announced that its decoupling technology for Prefill and Decode has successfully expanded from a single cluster to cross-data center and heterogeneous hardware environments. According to the article, this move is expected to significantly reduce the inference cost per token. Previously, the expansion of this technology was hindered by issues related to KV cache transmission overhead. The realization of this breakthrough depends crucially on its hybrid model Kimi