Monitoring inference cost

3.4Standard

A generative-AI app exceeds its monthly budget as usage grows.

Find the cause while maintaining quality.

Choose two appropriate monitoring activities.

Hint 1: Focus on the requirement

Break usage into measurable drivers and connect those drivers to owners or features before optimizing cost.

Hint 2: Key terms

Review the usage and allocation concepts behind inference cost monitoring.

Token usage: The number of input and output token units processed by model requests.
Request count: The number of calls made to an inference endpoint or model operation.
Cost allocation: Assigning spend to teams, users, features, models, or other accountable dimensions.
Budget: A spending limit or forecast used to detect and manage expected cost.

Question IDAWS-AIF-083

EnvironmentAWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認

Last technical review2026-08-20

More practice in this area