A setting that controls the randomness of an AI model's output — lower values produce more predictable responses, higher values produce more varied ones.
Temperature is one of the first parameters developers tune: near-zero for tasks needing consistency (like structured data extraction), higher for creative or brainstorming tasks.
