💨 Abstract

OpenAI's new o3 and o4-mini AI models are state-of-the-art but hallucinate more than older models, according to internal and third-party tests. These "reasoning" models hallucinate in response to 33% of questions on PersonQA, which is double the hallucination rate of previous reasoning models.

Courtesy: techcrunch.com

Summarized by Einstein Beta 🤖