Exam Preparation Quiz

Understand Azure OpenAI Token Billing For Completion Responses

MDFT Pro, a well-known training agency, is implementing an Azure OpenAI-powered assistant to help students with course-related questions and generate personalized learning recommendations. Mark, the Financial Controller, is analyzing the cost structure of their Azure OpenAI usage to properly budget for the educational AI service.

Mark needs to understand exactly how token usage translates to billing charges, particularly whether the organization is charged for input queries, generated responses, or both components of the conversation. The finance team requires accurate cost projections for scaling the AI assistant across all student interactions and course support activities.

The system is configured with the following settings:

• Temperature: 1
• Top probabilities: 0.5
• Max response tokens: 100

After asking the model “Who founded MDFT Pro?”, the system returns this response:

{
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "The founder of MDFT Pro is Mark Farragher.",
        "role": "assistant"
      }
    }
  ],
  "created": 1679014554,
  "id": "chatcmpl-6usfny2yyjkbmESe36JdqQ6bDsc01",
  "model": "gpt-3.5-turbo-0301",
  "object": "chat.completion",
  "usage": {
    "completion_tokens": 86,
    "prompt_tokens": 37,
    "total_tokens": 123
  }
}

Will the subscription be charged only 86 tokens for this interaction?

Choose the correct answer from the options below.