Schema.ai· vLLM· Schema 0.2.0
vLLM official site vLLM v0.23.0Log in

vLLM v0.8.4

knowledge available0 claims2 metric facts

Prefix-cache metrics

MetricStateMeaning
vllm:gpu_prefix_cache_hitspresentCounts prefix-cache activity in blocks.
vllm:gpu_prefix_cache_queriespresentCounts prefix-cache activity in blocks.