SESKStuff Everyone Should Know
SESK: Stuff Everyone Should Know · sesk.info/ai-temperature-randomness

What happens when you raise a language model's temperature setting?

Output gets more random. Lower temperature makes it more predictable.

Start playing free

  1. The model runs faster
  2. The output gets more random ✔
  3. The model gets more accurate

Why: Temperature reshapes the probabilities of the next token before one is picked. High values flatten them so unlikely tokens get chosen more often; low values make the top choice win almost every time.

Huh, didn't know that: Temperature is a hyperparameter you set when generating, not something the model learns.

The dad joke
I set my chatbot's temperature to high. Now it writes poems about my toaster.
Like this card?
Play the game

References

Answer: The output gets more random. Temperature reshapes the probabilities of the next token before one is picked. High values flatten them so unlikely tokens get chosen more often; low values make the top choice win almost every time.

  1. Google Machine Learning Glossary: Generative AI (temperature) developers.google.com/machine-learning/glossary/generative
    Higher temperatures result in more random output, while lower temperatures result in less random output.
    Checked 2026-10-10.

Think this answer is wrong? Tap "Challenge this answer" on the card in the game and tell us why.

Test yourself in the game

Artificial IntelligenceHow can an unrepresentative training sample bias an AI system?Artificial IntelligenceHow do AI, machine learning and deep learning fit together?Artificial IntelligenceWhat did Alan Turing call his 1950 test of machine intelligence?

That's stuff everyone should know.