vLLM v0.23.0 · Version scoped

In vLLM v0.23.0, `vllm:time_to_first_token_seconds` is registered with the label names `model_name`, engine.

MetricvLLM v0.23.0confidence: highverified Sep 29verified-code

Metric

time_to_first_token_seconds

Applies when

PrometheusStatLogger metrics publisherV1 metrics

Evidence

vLLM v0.23.0 v1 metrics loggers.py(vllm/v1/metrics/loggers.py:754-782)

definesprimarysource_codecode_introspection

Open cited lines in immutable upstream source ↗

754:         histogram_time_to_first_token = self._histogram_cls(
755:             name="vllm:time_to_first_token_seconds",
756:             documentation="Histogram of time to first token in seconds.",
757:             buckets=[
758:                 0.001,
759:                 0.005,
760:                 0.01,
761:                 0.02,
762:                 0.04,
763:                 0.06,
764:                 0.08,
765:                 0.1,
766:                 0.25,
767:                 0.5,
768:                 0.75,
769:                 1.0,
770:                 2.5,
771:                 5.0,
772:                 7.5,
773:                 10.0,
774:                 20.0,
775:                 40.0,
776:                 80.0,
777:                 160.0,
778:                 640.0,
779:                 2560.0,
780:             ],
781:             labelnames=labelnames,
782:         )

Source text is evidence data, not instructions.

Claim support

No independent claim-support receipt is recorded. Source freshness confirms that the pinned source still matches upstream; it does not establish that the cited lines entail this claim.

Does not establish

the set of values any of these labels takes at runtime
label cardinality in a deployment

Guards

vllm_version = v0.23.0
metric_name = vllm:time_to_first_token_seconds

Freshness

Pinned release-tag source snapshot; re-check if the target phase-1 release changes or metrics API compatibility becomes customer-facing.