A customer chat service must return short answers with high concurrency.
Response time and cost matter more than complex long-form reasoning.
Choose two model-selection criteria.
Compare measured latency and throughput for representative prompts, along with inference cost per request or token.
Detailed explanation
A file name does not characterize a model's service behavior.
At scale, token volume and unit price directly affect operating cost.
The service must meet response-time and concurrency requirements with representative inputs.
A name's length says nothing about capability, latency, or price.
Check official documentation
Compare the following points with the official documentation.
Amazon Bedrock公式のモデル・推論料金・サービスクォータ情報を確認し、代表プロンプトで測定する。
Official references
What to confirm
レイテンシ・スループット・コストをモデル選択の要件として比較できる。
Key points
- Latency
- Throughput
- Inference cost
Notes
- Environment: AWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認
- Compare the definition with the AIF-C01 objective and the service documentation before applying it.
- Keep the concept separate from a particular answer choice or implementation step.
Foundation review
Tail latency
A percentile such as p95 shows slower requests that an average can hide.
Capacity trade-off
Higher throughput or lower latency can require different model, batching, or infrastructure choices.