inference/research
verifiedhigh · 0.94updated 2026-10-04

vLLM prefill time-to-first-token histogram

vllm · histogram

Fields

id
metric-exposures/w3m-vllm-vllm-time-to-first-token-seconds
Metric name
vllm:time_to_first_token_seconds
Metric type
histogram
Unit
seconds
Exposure
prometheus
Endpoint
:8000/metrics
Histogram buckets
0.0010.0050.010.020.040.060.080.10.250.50.7512.557.5102040801606402560
Labels
model_nameengine
Enabled by default
yes
Phase scoped
yes
notes
vLLM main, PrometheusStatLogger in vllm/v1/metrics/loggers.py, read 2026-10-04. On by default (EngineArgs.disable_log_stats defaults False); served at :8000/metrics. Bucket family name in --custom-histogram-buckets: time_to_first_token. Interval from request arrival to first output token. Same label set and same unit as every other vLLM metric, so a TTFT SLO can be defended on documented buckets.

Notes

vLLM main, PrometheusStatLogger in vllm/v1/metrics/loggers.py, read 2026-10-04. On by default (EngineArgs.disable_log_stats defaults False); served at :8000/metrics. Bucket family name in --custom-histogram-buckets: time_to_first_token. Interval from request arrival to first output token. Same label set and same unit as every other vLLM metric, so a TTFT SLO can be defended on documented buckets.

Cross-references