AI Dictionary › AI Fundamentals
Temperature is a parameter that controls the degree of "randomness" or "creativity" in a language model's responses. It is usually expressed on a scale from 0 to 1 (or up to 2 in some models). At low temperature (near 0), the model almost always chooses the statistically most likely token, producing more consistent, predictable, and conservative responses. At high temperature (near 1 or above), the model diversifies its choices more, producing more creative, varied but potentially less accurate outputs.
The parameter that scales randomness in token choice: low values make the model pick the most likely words (consistent, repetitive), high values widen the choice (varied, riskier). Ranges differ by provider, but the intuition holds everywhere.
Extraction, classification, JSON output: temperature 0. Marketing copy variants: 0.8-1.0 and generate five options. Same prompt, different tool depending on the dial.
Set it deliberately per task instead of leaving the default. And know its limit: temperature 0 reduces variance but does NOT make outputs deterministic: same prompt can still produce different answers. If you need repeatability, measure it: run the prompt N times in the Bench and count.
Multi-turn prompting · JSON mode · Prompt scaffolding · Contrastive prompting · Context stuffing · Retry with backoff · Self-ask · Chain-of-thought · Role prompting · Prompt injection · System prompt
From our network
AGORÀ Intelligence: Enterprise AI Governance Platform
Govern AI at scale: policies, adoption and measurable results on your data. Built for boards and C-suite.
Visit agora-intelligence.com →From the Agora Intelligence blog
📱 Download the Android app (beta) iOS coming soon
Say what you mean. Get what you need.
Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.