Skip to content

Temperature

A setting that controls how random or creative a model's output is.

Last updated

Temperature scales the randomness of token selection. A low value makes output more predictable and factual, which suits extraction and coding. A higher value produces more varied and creative text. Most APIs expose it alongside related controls such as top-p, and it is one of the first knobs to tune when results feel either too rigid or too erratic.

Related terms

Keep exploring

Browse the full AI glossary or compare AI tools that use this technology.

Report an issue with this page