GPT-5.1 is smarter and more conversational
OpenAI is releasing GPT-5.1, which brings improvements to the intelligence and the communication style of ChatGPT. The new models include GPT-5.1 Instant and GPT-5.1 Thinking.
The company says that GPT-5.1 Instant is “warmer, more intelligent, and better at following your instructions” than its predecessor. The advanced reasoning model GPT-5.1 Thinking is working faster on simple tasks and is more persistent on complex ones, while it’s easier to understand.
Answers from the GPT-5.1 Instant model appear more natural in the examples shared by OpenAI. The model is also better at following instructions and can decide to think before responding to more challenging queries. The GPT-5.1 Thinking model sounds more approachable and gives more empathetic and warm answers.
More personality options
You will be able to access the older GPT-5 models from the legacy menu for three months. | Image credit – OpenAI
A major part of the update is the expanded personality presets for the conversational tone of the models. Users will be able to choose from eight options for the tone of how ChatGPT responds. Those are:
- Default
- Professional
- Friendly
- Candid
- Quirky
- Efficient
- Nerdy
- Cynical
OpenAI is also going to debut an experimental new way to fine-tune the style of ChatGPT. Some users will start being able to tune ChatGPT’s characteristics directly from the personalization settings. They will be able to select how concise, warm, or scannable its responses are, and how frequently it uses emojis.
The new models are rolling out today, starting with paying users. Free and signed-out ChatGPT users will get access after that. Unlike the abrupt disappearance of GPT-4o in August, the old GPT-5 models will be accessible for three months from the ChatGPT legacy models dropdown menu.
What about hallucinations?
While I appreciate the opportunity to customize ChatGPT, the style of its answers has never been much of an issue for me. I’d rather OpenAI focus on minimizing hallucinations and improving the accuracy of ChatGPT and its models. On the other hand, I don’t think I’ll stop double-checking almost everything an AI tells me anytime soon.
