Does a temperature of zero give an identical result every time?
Answer out loud or in your head first, then compare.
Answer
Don't treat it as a general guarantee. Besides the token-selection parameter, the model, its version and how the provider runs it all matter. Not every model exposes the same set of parameters either. Less variation doesn't mean truthfulness.
In tests I record the settings and check behaviour, not always the literal same text. Two differently worded answers can be equally correct. When a repeatable business decision matters, I move clear-cut rules into code or a required approval. After changing the model, I evaluate the results again, even if the prompt and temperature stayed the same.
Why it matters: Helps avoid a false assumption about the repeatability of AI tests.
Possible follow-ups: When does identical text make sense as a test, and when not?
How did it go? Be honest — “Not yet” puts the question on your review list.
Start with
- F-P01 · What is an LLM? — LLM fundamentals and AI engineering
- F-P03 · What are a token and a context window? — LLM fundamentals and AI engineering
Further reading
- Claude Platform: Define success criteria and build evaluations(opens in a new tab)Docs; accessed 22.10.2026. Success criteria, representative data and kinds of graders.