OpenAI’s latest iteration of its AI chatbot, ChatGPT GPT-5.1, now comes in two new AI models with improvements in reasoning, instruction-following, clarity, and tone.
The most used model, GPT-5.1 Instant, is more conversational with a warmer, more intelligent tone than an earlier chat model. It is also said to be better at following instructions and has an adaptive reasoning capability that lets it decide when to think before responding.
The more advanced reasoning model, GPT-5.1 Thinking, adapts thinking time more precisely to each question. It is now easier to understand and faster on simple tasks, and more persistent on complex ones, according to OpenAI.
GPT-5.1 has an automatic routing system that analyses each request and will direct it to the model it is best suited for, so that in most cases the user does not need to decide on the model to use.
These chat models are trained using the same stack as OpenAI’s reasoning models, so they score higher on factuality and complex problem-solving than GPT-5, while also introducing a more natural, conversational tone.
Both models will begin rolling out to users this week, with application programming interface (API) access to follow. GPT-5 (Instant, Thinking, and Pro) will still be available under the legacy models dropdown for paid subscribers for three months.
The company has also increased the number of personality presets, that include Default, Friendly, Efficient, Professional, Candid, Quirky. In addition, there is an experiment for new ways to fine-tune ChatGPT’s style directly from settings. OpenAI will later provide an update of GPT-5 Pro to GPT-5.1 Pro.
Fidji Simo, chief executive officer for applications at OpenAI, said in a blogpost that the company’s research has found that “people experience ChatGPT in highly individual ways.” Where some prefer direct and neutral responses, others want empathy and conversation, she noted.
“They want ChatGPT to feel like a single assistant that appropriately adapts its tone to the context of the conversation,” she added.
An example is for the chatbot to display empathy when the conversation is about health or relationships, but be more direct on search and copywriting.
“In the past, building for global scale meant creating the most universal product possible [and]… a consistent user experience,” she said. “However, that is not true for model personality, which people prefer to be more tailored to their preferences.”
