Temperature Playground

See how temperature affects token selection. Generate text, then click on any token to see the probability distribution the model used when choosing it.

How Temperature Works

At each step, the model assigns probabilities to possible next tokens. Temperature controls how "peaked" this distribution is. Click any token after generation to see what alternatives the model considered.

T=0: Always picks the highest probability token
T=0.7: Usually picks likely tokens, occasionally surprises
T=1.5: Lower probability tokens have a real chance

Prompt

Try an example:

Compare temperatures (select 1-3):

T = 0.0Precise

Deterministic, always picks the most likely token

Output will appear here after generation

T = 0.7Creative

More diverse, interesting outputs

Output will appear here after generation

T = 1.5Chaotic

Very high variance, surprising results

Output will appear here after generation

Temperature Scale Reference

0.0

0.5

1.0

1.5

2.0

Deterministic / FocusedBalancedCreative / Random