Temperature is a setting that controls how random or creative a language model’s output is. A low temperature makes responses focused and predictable; a high temperature makes them more varied and creative.
What it means in plain English
When a model generates text, at each step it has a range of possible next words with different probabilities. Temperature adjusts how strictly it sticks to the most likely choice. Near zero, it almost always picks the single most probable word, giving consistent, deterministic answers. Higher temperatures let it choose less likely words more often, producing more diverse and surprising output — at the cost of some reliability.
It is a dial for the trade-off between consistency and creativity.
A simple example
For a factual question where you want the same reliable answer every time, a low temperature is best. For brainstorming creative story ideas, a higher temperature produces more varied and imaginative suggestions.
Why it matters
Temperature is one of the main levers for shaping AI output to a task. Understanding it explains why the same prompt can give identical answers or wildly different ones, and lets you tune a model toward precision or creativity as needed.
Related terms
- Large Language Model — what the temperature setting applies to.
- Prompt — the input temperature affects the response to.
- Inference — when temperature takes effect.
Frequently asked questions
What does temperature control in AI?
Temperature controls the randomness of a language model’s output: lower values make responses more focused and deterministic, higher values make them more varied and creative.
What temperature should I use?
Lower (near 0) suits factual, consistent tasks; higher suits brainstorming and creative writing. The best value depends on how much variety you want.
