Temperature
1 min read · Reviewed 17 Oct 2026
In short
A dial between predictable and inventive.
In plain English
Temperature controls how much the model varies its choices. Low temperature makes it pick the most likely next word almost every time — consistent, sometimes flat. Higher temperature lets it take less obvious paths.
A real example
For extracting data from an invoice you want it near zero: the same input should give the same output. For brainstorming names you want it higher, or you get the obvious answer every time.
Why it matters
If a tool gives inconsistent answers to identical questions, temperature is usually why. Where an API exposes it, it is one of the few settings genuinely worth adjusting.
