OpenAI's new o3 and o4-mini models hallucinate more than the company's predecessor reasoning models.

OpenAI is facing a worrying problem with its latest models of AIA reasoning, o3 and o4-mini. While these new models were supposed to represent a significant advance in artificial intelligence, they turn out to be more subject to hallucinations than their predecessors — a phenomenon that raises many concerns within the industry. [...] The article Hallucinations: OpenAI recognizes a problem with o3 and o4-mini appeared first on BlogNT: the Blog of…

OpenAI's new o3 and o4-mini reasoning models offer more results with hallucinations than the company's previous reasoning models, with almost twice the hallucination rate recorded in model o1, according to internal evidence from the PersonQA evaluation.

PandaFiredoge compiled a timeline of the release of different OpenAI models to show their development history. Due to the rapid development of AI technology, this timeline may soon become outdated.