Inference performance

Model replica

Also known as Model replica, AI Model replica

By WeavePublished 1 min read

Definition

Model replica is the serving concept concerned with model replica during AI inference.

What model replica means

Model replica is the serving concept concerned with model replica 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

An engineer can isolate this behavior in a warm serving test before comparing complete user tasks. 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 model replica without evidence.

Limitations and tradeoffs

A benchmark that omits cold starts, cache state, retries, or queueing may not represent production traffic. 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 model replica beside model, provider, workload, and request outcome. That context helps explain whether a routing change altered model replica, 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

Sources and further reading

  1. OpenTelemetry GenAI semantic conventions