The now penultimate major update of ChatGPT arrived last August amid much expectation that she was not satisfied. User rejection of the language model GPT-5 which replaced GPT-4o led the company to react quickly and recover it as another option within the chatbot. This Wednesday, OpenAI has begun the global deployment of the new LLM which takes the place of GPT-5: GPT-5.1.
Like the previous model, GPT-5.1 arrives under the variants instant and thinking. The first is a more conventional model, good at following instructions, but now designed to be ‘warmer’ and smarter. In general terms, GPT-5.1 Instantis smarter, faster, expressive and conversational.
For its part, GPT-5.1 Thinking is an advanced model of reasoning that has been renewed to understand and solve simple tasks more quickly and be more consistent when faced with complex tasks. OpenAI says that GPT-5.1 Thinking is globally better than the previous version of advanced reasoning and also more accessible, as uses less jargon and is more empathetic and warm by default.
With GPT-5.1, OpenAI offers new customization options in ChatGPT to adjust the style of the model to that of the user. Furthermore, it incorporates a adaptive reasoningimproved which allows you to decide when you should ‘think’ before answering complex questions, improving your performance on several assessment metrics.
OpenAI states that Both models perform better on technical tests such as math and coding assessments (including AIME 2025 and Codeforces) than GPT-5.
As was the case with GPT-5, GPT-5.1 Auto will decide which model to direct each request to when the user does not expressly select one. Regarding its availability, the new model is being gradually rolling out to ChatGPT Pro, Plus, Go and Business users, followed later by free users. It will also be available in the API later this week.
Retirement of previous chatbot models
Regarding the withdrawal of the previous models, it is not immediate as it happened last time, but it is close. The company has said that ‘GPT-5 (Instant and Thinking) will remain available on ChatGPT within the legacy models menu for paying subscribers for three monthsso they can compare and adapt at their own pace. The GPT-5 withdrawal period does not affect other legacy models. Going forward, when we introduce new ChatGPT models, our focus will be to provide enough room to evaluate changes and share feedback, allowing for further innovation on more advanced models and facilitating a smooth transition. These periods will be communicated clearly and in advance.’ That is to say, GPT-4o will still be available, although GPT-5 will disappear.
The 8 personalities of ChatGPT
Regarding the new customization options in ChatGPT, OpenAI now allows users to choose between 8 conversation tones or chatbot personalities to better fit the user’s needs, 3 more than what it offered until now:
- Default: balanced style and tone.
- Professional: polished and precise.
- Gentle: warm and close.
- Sincere: direct and motivating.
- Peculiar: fun and imaginative.
- Efficient: concise and clear.
- Geek: explorer and enthusiast.
- Cynical: critical and sarcastic.
The new options are Professional, Sincere and Quirkybut the others have also received updates. All of these predefined personality options work injecting different instructions into the system request that the model processes before generating each response.
ChatGPT will soon be able to ask if you want to update your tone preferences when you detect repeated requests in a different style.
ChatGPT already has 800 million users
Fiji SimoCEO of applications at OpenAI, notes in a blog post that the company wants ChatGPT to ‘adapt to your needs and work in the way that best suits you.’ Simo explained that, with over 800 million ChatGPT usersthe company has moved beyond standardized approaches. He added that each user’s experience with ChatGPT is different: some prefer direct and neutral responses, while others opt for different response patterns.
The figure of 800 million users is a new ChatGPT recordgiven that no other application has reached such a volume of users in the period of time since OpenAI launched the chatbot to the general public in November 2022. Simo does not specify the frequency of these users, but given that the last figure the company gave was 700 million weekly usersthese should be too.