Event ·Multi-source

Kimi K3 发布，KDA 前缀缓存贡献给 vLLM

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。

Current read

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。

11 sources83° AI heatLast updated 2026/7/18 12:25:38

Evidence

11 sources

Primary sourcevLLM

2026/7/18 12:25:38View source

Related source 1lmarena.ai

2026/7/16 20:41:40View source

Related source 2The Rundown AI

2026/7/16 18:46:16View source

Related source 3AI SDK

2026/7/16 19:52:19View source

Related source 4elvis

2026/7/16 21:24:38View source

Related source 5marktechpost

2026/7/16 23:47:05View source

Related source 6pandaily

2026/7/17 02:02:16View source

Related source 7orange.ai

2026/7/17 02:22:12View source

Related source 8Decoder

2026/7/17 19:16:07View source

Related source 9@koltregaskes

2026/7/19 12:06:44View source

Related source 10SuperTechFans

2026/7/19 22:39:41View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story