精选理由
Kimi 和 vLLM 合作搞了个新缓存方案，K3 模型上来就能高效处理长文本，开源权重明年7月放出来。
Kimi Moonshot 发布新模型 Kimi K3，并将 KDA 前缀缓存实现直接贡献给 vLLM。KDA 突破了传统前缀缓存的假设，可提升长上下文服务的效率。vLLM 将在发布当天（July 27, 2026）开放权重时同步支持 K3 模型。该合作确保了社区从第一天起就能高效使用长上下文推理。
原文 · vLLM
Congrats to @Kimi_Moonshot on the Kimi K3 announce…
Congrats to @Kimi_Moonshot on the Kimi K3 announcement! 🎉
Grateful for the shoutout and collab. The Kimi team announced they contributed a KDA prefix caching implementation directly to vLLM, to be released alongside the model. 🚀
KDA breaks assumptions behind conventional prefix caching, and this upstream work means the community gets efficient long-context serving from day 0.
vLLM will support Kimi K3 on day-0 release.
Open weights by July 27, 2026.