vLLM(@vllm_project):Kimi K3 serving in vLLM now delivers 2.2–2.8x throughput on our B300 benchmark vs v0.27.1. We break down the work across scheduling, KDA state handling, and MoE kernels, with benchmarks and commands to reproduce the results. Thanks to the vLLM community for pushing Kimi K3 performance forward! Read the deep dive:

vLLM

@vllm_project

A high-throughput and memory-efficient inference and serving engine for LLMs. Join to discuss together with the community!

加入 March 2024

36 正在關注 50.2K 粉絲

vLLM@vllm_project

2026.09.17 15:57

Kimi K3 serving in vLLM now delivers 2.2–2.8x throughput on our B300 benchmark vs v0.27.1. We break down the work across scheduling, KDA state handling, and MoE kernels, with benchmarks and commands to reproduce the results. Thanks to the vLLM community for pushing Kimi K3 performance forward! Read the deep dive: