Temperature is a parameter that controls how predictable or varied an AI model's outputs are. A low temperature produces more focused and consistent responses, favoring the highest-probability outputs. A higher temperature introduces more variation — sometimes producing more creative outputs, sometimes less reliable ones.
In enterprise settings, low temperature is typically appropriate for tasks requiring consistency and accuracy — classification, extraction, structured generation. Higher temperature might be appropriate for brainstorming, creative content generation, or generating diverse options for human review. Getting temperature right for each use case is a small but meaningful tuning decision.
