OpenAI claims ChatGPT’s new default model hallucinates way less

OpenAI has launched its new GPT-5.5 Instant model, showcasing significant improvements in factual accuracy. According to internal evaluations, the new model produced 52.5% fewer hallucinated claims compared to GPT-5.3, particularly in high-stakes areas like medicine, law, and finance. These enhancements come at a time when AI faces challenges with hallucinations, marking a crucial step toward reducing errors in complex conversations. The results also indicated a 37.3% reduction in inaccurate claims flagged by users.