OpenAI has started rolling out GPT-5.1, the next big update to its ChatGPT models — designed to make conversations feel more natural, thoughtful, and personal.
The release includes two upgraded versions: GPT-5.1 Instant and GPT-5.1 Thinking.
The Instant model, which most people use daily, now sounds warmer and more conversational while improving at following instructions. It also learns when to “think” before replying to complex questions, helping it give more accurate answers without slowing down simple chats.
The Thinking model focuses on deep reasoning and problem-solving. It’s now faster on easy tasks but takes extra time on harder ones, producing clearer, more structured answers. OpenAI says it’s also easier to understand, even when explaining complex technical ideas.
Alongside these upgrades, OpenAI is giving users more control over ChatGPT’s personality. You can now choose tones like Professional, Candid, or Quirky, or even fine-tune how concise, warm, or detailed the responses are. These preferences can update instantly, even mid-conversation.
GPT-5.1 is rolling out this week to paid users first, with free users getting access soon after. Both models will also appear in the API later this week.
OpenAI says GPT-5.1 marks a step toward making ChatGPT not just smarter, but more enjoyable to talk to — an AI that feels a little more like you.
What makes GPT-5.1 different?
GPT-5.1 aims to enhance the naturalness and intelligence of conversations. OpenAI characterizes this update as an improvement for both its Instant and Thinking models. The GPT-5.1 Instant model, which most users interact with, is now described as “warmer, more intelligent, and better at following your instructions.
FAQs
What’s new in GPT-5.1?
GPT-5.1 brings a warmer tone, more intelligent reasoning, and better instruction-following, with improved versions of both the Instant and Thinking models.
Who gets access to GPT-5.1 first?
Paid ChatGPT users on Pro, Plus, Go, and Business plans are receiving GPT-5.1 first, with free users to follow soon.
