Temperature
A sampling parameter (0-2) that controls how random a model's output is.
Temperature scales the probability distribution from which the model samples its next token. At temperature 0 the model always picks the most likely next token (deterministic, repetitive). At higher temperatures, less likely tokens become possible (more creative, more random).
Typical settings: 0 for code, classification, and structured output; 0.3-0.7 for assistants and chat; 0.8-1.2 for creative writing and brainstorming.
Temperature is one of two main sampling controls; top-p (nucleus sampling) is the other. Most APIs let you set either, but you should usually pick one and leave the other at default.