Skip to content
inference.academy

Benchmarks

Each report sends the same controlled requests down every route that serves a model, and publishes the per-call data next to the charts, so any number here can be checked or recomputed. The next step is replaying a workload you choose against those measurements.