Skip to content

Evaluation

GPT-4 internal factuality evaluations (closed- and open-domain hallucinations)

closed-domain and open-domain hallucination rates

Hallucination & factual accuracy OpenAI

How it was run

Open a release to see the setup, scoring, and other details that the lab published. Only fields the lab actually disclosed are shown.

GPT-4
OpenAI

rich detail 10 method fields published

See how this test was run