What is temperature?

By the Variask team · Updated

Definition

Temperature is a setting that controls how random an AI model's output is. Higher temperature gives more varied answers, lower gives more predictable ones. It is one reason the same question can get different answers.

Also called Model temperature

How it works

  • Models pick each next word from likely options.
  • Low temperature favours the most likely words.
  • High temperature allows less likely words more often.
  • Consumer apps set temperature themselves, users usually cannot.

What it means for brands

  • Some variation in answers is built in. Measure across many runs.
  • Brands named consistently at any setting are the strongest.

Questions

Temperature:
common questions

Not in the consumer app. Developers can set it through the API.

It can, at the margin. Brands with strong, consistent sources are named regardless.

Related terms

Keep exploring

Measurement

Answer volatility

Answer volatility is how much AI answers to the same question change from one run to the next, or from week to week. High volatility means a single check can be misleading, so scores need many answers and a likely range.

Read

How AI search works

Large language model (LLM)

A large language model (LLM) is an AI model trained on very large amounts of text to understand and write language. LLMs power assistants such as ChatGPT, Gemini, Claude, Grok and DeepSeek.

Read

How AI search works

System prompt

A system prompt is the hidden instruction an AI product gives its model before the user's question, setting its role, tone and rules. It shapes how answers are written, such as whether to name brands or cite sources, but users do not see it.

Read
Full glossary