In vLLM v0.23.0, `vllm:request_prefill_kv_computed_tokens` is computed in finished-request logging after request completion from `num_prompt_tokens` and `num_cached_tokens`.

MetricvLLM v0.23.0confidence: highverified 11 hr agoverified-code

PrometheusStatLoggerV1 metrics

vLLM v0.23.0 v1 metrics loggers.py(vllm/v1/metrics/loggers.py:1179-1213)

definesprimarysource_codecode_introspection

Pinned release-tag semantics; do not carry to another version without continuity evidence.

In vLLM v0.23.0, vllm:request_prefill_kv_computed_tokens is computed in finished-request logging after request completion from num_prompt_tokens and num_cached_tokens.