ChatGPT's New Model Hallucinates 52% Less

OpenAI claims its new default ChatGPT model, GPT-5.5 Instant, significantly reduces hallucinations. The company says internal evaluations show the model produces 52.5% fewer hallucinated claims than its predecessor. Hallucinations, where AI models generate false or fabricated information, have been a persistent problem across the industry. OpenAI describes the improvement as significant gains in factuality across the board.