ChatGPT's Default AI Model Upgraded to GPT-5.5 Instant: Hallucinations Reduced by 52.5% at Most

Many users worry about false information (hallucinations) in AI chats affecting decisions. OpenAI recently announced replacing ChatGPT's default model with GPT-5.5 Instant, focusing on accuracy and conciseness while cutting unnecessary emojis. In high-risk scenarios like healthcare, law and finance, OpenAI's internal evaluation shows hallucinations in GPT-5.5 Instant dropped by 52.5% compared to the previous version, enhancing the reliability of AI responses.