Tail latency
Also known as Tail latency, AI Tail latency
Definition
Tail latency is the serving concept concerned with tail latency during AI inference.
What tail latency means
Tail latency is the serving concept concerned with tail latency during AI inference. In a serving system, define the measurement boundary before collecting data. State whether the worker is warm, which model and precision are active, how batching is configured, and whether the request is streamed. These details determine whether two similar measurements describe the same operating condition.
How it appears in a workload
A gateway trace can connect tail latency with the selected model, provider, queue, and completion outcome. Record relevant timestamps and runtime signals with the request rather than relying on one dashboard aggregate. A useful comparison includes request mix, arrival pattern, concurrency, and output limits. For routed traffic, retain the selected route so a change can be investigated rather than attributed to tail latency without evidence.
Limitations and tradeoffs
This term describes one serving property, not model quality or a promise that every request behaves identically. Use it with p50 and tail latency, useful token throughput, errors, memory pressure, and cost. The best configuration is the one that meets the application's reliability and responsiveness requirements under representative traffic.
How this relates to Weave
Weave Router comparisons can record tail latency beside model, provider, workload, and request outcome. That context helps explain whether a routing change altered tail latency, but Weave does not make an inference runtime faster by itself. Keep serving conditions visible when using these observations to compare complete AI tasks.
Explore Router