When an LLM confidently states facts that are false or fabricates citations. The gap between output plausibility and output correctness.
reliability
Hallucination
Related terms
Tools that implement Hallucination
LangSmith
Evaluation · Platform (any LLM)
8.7
LangChain's eval + observability platform.
Freemium· Free starter; Plus $39/mo per seatLLM tracingevals
Braintrust
FeaturedEvaluation · Platform (any LLM)
8.9
Eval, monitor, and improve AI products end-to-end.
Freemium· Free up to 1k events/day; team from $249/moevalsmonitoring