Question

A development team is deploying a large language model on Azure to power an internal document assistant. During cost planning, the finance manager asks how the model's usage will primarily be measured for billing. What is the correct explanation the team should give?

Accepted Answer

Usage is measured by the number of tokens processed in both the prompt (input) and the completion (output).

Answer

Usage is measured only by the number of separate API calls made, regardless of message length.

Answer

Usage is measured by the number of concurrent users connected to the assistant at any time.

Answer

Usage is measured by the number of documents stored in the connected knowledge base.