RHEL AI uses vLLM for model serving and it provides a /metrics endpoint with information described below.  A sample Grafana dashboard is linked below also, which may be a good candidate for grafana-pcp.  Upstream PCP PR #2066 enables these metrics by default when available (should arrive via rebase to pcp-6.3.1 in a week or so).

https://docs.vllm.ai/en/latest/serving/metrics.html

https://github.com/vllm-project/vllm/tree/main/examples/production_monitoring

https://github.com/performancecopilot/pcp/pull/2066