GPT-5.1 is here: "more intelligent and entertaining"

With GPT-5.1 and GPT-5.1 Thinking, two new AI models are joining ChatGPT. OpenAI describes them primarily as “more intelligent and entertaining,” and the question of what is actually intelligent has arisen not only with the advent of generative AI. The new versions “bring improvements in logical reasoning, instruction following, clarity, and tone.” However, ChatGPT is also intended to become more personal and customizable upon request.

The AI models are sophisticated. Specifically, there is a GPT-5.1 Instant that can follow instructions better. GPT-5.1 Thinking is, according to the blog post, OpenAI's most advanced reasoning model. It is intended to be easier to understand, thus being able to process content well for non-experts, as well as being faster at simple tasks and more persistent at complex tasks.

GPT-5.1 Auto selects the appropriate model for each request. Therefore, when asking ChatGPT a question, you generally don't have to check which model can best answer the question. The chatbot recognizes complex questions that require in-depth research, just as it knows when only a quick query for information is sufficient.

Smooth transition to GPT-5.1

The previous GPT-5 Pro will be automatically updated to GPT-5.1. Here too, the user doesn't have to do anything. The new models will only be available via the API according to OpenAI's blog post shortly. After there were some protests from users during the last switch to GPT-5, OpenAI is offering a smoother transition this time for paying subscribers. They can still access GPT-5 for three months. The old model can be selected from the drop-down menu.

The displeasure of some users at the time was related to the rather cooler tone that GPT-5 adopted compared to its predecessor GPT-4o. OpenAI is aware that there are people who believe they are forming a close relationship with the chatbot. They are working to figure out the best way to handle it. “As this technology evolves and new applications emerge, many new challenges will arise. In a project of this magnitude, we must never assume that we have all the answers.”

The preset personalities that already exist for ChatGPT are being expanded. In addition to the styles Standard, Friendly, Efficient, Professional, Sincere, Quirky, there will also be, for example, nerdy, cynical, and outspoken. What's new is that you can fine-tune the style directly via the settings in the future; this is expected to work much better than before. Fidji Simo, CEO of Applications at OpenAI, explained in a blog post why she considers further personalization of AI important. “We want ChatGPT to feel like yours.” With 800 million users, a “one-size fits-all” solution can no longer fit.

(emw)