# huawei-csl/KVarN

KVarN is a native vLLM KV-cache quantization backend for your agents: 3-5x more context, throughput above FP16, and FP16-level accuracy. Calibration-free, one flag.

**Commercial license**：Commercial OK — 可商用，通常只需保留著作權聲明/授權條款

**Stars**：440
**Source**：https://github.com/huawei-csl/KVarN