What is temperature in AI?
The setting that decides whether an AI plays it safe or gets creative.
Ask an AI the same question twice and you may get two different answers. One reason is a setting called temperature.
A setting that controls how predictable or varied a model’s answers are.
A chef. At a low temperature, they follow the recipe exactly, every time. At a high temperature, they start improvising: sometimes brilliant, sometimes strange.
What it actually changes
Every time a model writes the next word, it has a list of candidates, each with a likelihood.
Which setting for which job
- LowFacts, code, support replies
- MediumEveryday writing
- HighBrainstorming, stories, names
Most tools use a scale from 0 to 1, and some go up to 2. Many chat apps set it for you; you usually adjust it yourself only when using AI through an API or a playground.
Why it matters
- Consistency. For tasks that need the same answer every time, lower is better.
- Creativity. For fresh ideas, a little randomness helps.
- Risk. Very high settings can produce rambling or nonsense.
A common misunderstanding
“Temperature 0 means always correct.” It only makes answers more consistent, not more accurate. A model can be consistently wrong. See hallucination.
Two customers who ask a bank’s assistant the same question about fees should get the same answer. Apps like that keep the temperature low. A tool that suggests marketing slogans sets it higher, because variety is the point.
Related buzzwords
Prompt · Hallucination · Token