A nondeterministic system is not guaranteed to produce the same output each time it receives the same input. Many large language model systems are nondeterministic because sampling, model settings, or infrastructure behavior can vary across runs.
A system that is not guaranteed to return the same output for a given input. LLMs are generally nondeterministic; that is, LLMs typically generate different responses to the same prompt. Nondeterministic systems are generally much harder to test than deterministic systems. See also probabilistic.