AI-900Domain 5: Describe Features of Generative AI Workloads on AzureGenerative AImediumSingle choice
What does the temperature parameter control in generative AI responses?
A.The maximum response length
That's max tokens.
B.Response randomness — lower is more deterministic, higher is more varied/creative
Randomness/creativity dial. Correct.
C.The model's knowledge cutoff
Knowledge cutoff is baked into training — not a parameter.
D.API cost per request
Cost follows token usage — temperature doesn't price requests.
Why this answer
Temperature scales output randomness: ~0 = focused/deterministic (good for factual tasks), higher = more diverse/creative. Max tokens controls length; top-p is a related sampling control.
Correct answer: B · Last reviewed 2026-02-14
Official AWS references
Practice this interactively
Answer without seeing the solution first in the Generative AI practice session.
Practice Generative AI