Cost & Latency Index

Definition & methodology

Methodology

Sample methodology: median response latency and per-million-token cost under standard load.

Dataset

Sample workload of representative prompts of varying length.

Metric

composite index (lower is better) (lower is better)

Version

1.0

Last updated

2026-09-01

Source

https://example.org/benchmarks/cost-latency-index

Limitations

Sample limitation note: pricing changes frequently and this figure may be stale.

Model results