GPT 5.1

GPT-5.1 focuses on both capability and usability. Instant keeps chats fast but can decide when to think before answering, which improves math and coding results without losing speed. Thinking balances depth and latency by varying its internal reasoning time and uses plainer language for complex topics. New tone controls and stronger adherence to custom instructions make it easier to shape how ChatGPT responds. Rollout begins with paid users and expands shortly after, GPT-5.1 Auto routes queries to the best model, and the API adds gpt-5.1-chat-latest for Instant plus GPT-5.1 for Thinking. Over time GPT-5.1 becomes the default model.