By skyforbes 
ChatGPT just got another upgrade, and this time, OpenAI swears it is different. The company released GPT-5.1 today as a ChatGPT that provides a “smarter, more conversational” experience across the board.
As a step forward in capability and usability across ChatGPT models, GPT-5.1 comes in two variants: GPT 5.1 Instant and GPT 5.1 Thinking. Per the company, it is warmer, more intelligent and designed to handle everyday questions and tasks with more personality. The model now uses “adaptive reasoning,” which means that it can figure out how much reasoning power it should devote to a prompt. This way, it provides more thorough and accurate answers without losing the element of speed.
GPT-5.1 Thinking, on the other hand, offers more of advanced reasoning and clarity. According to OpenAI, the model can dynamically decide how much cognitive effort a question deserves. So, it gets to spend less time on easy prompts and more on challenging, multi-step reasoning. “Thinking” is also built to answer with “less jargon” and be better at explaining technical concepts in plain English. To top it off, there is a warmer and more empathetic layer to it.
The theme for this update seems to be the balance of IQ and EQ, as early testers reportedly describe the model as playful and precise.
“We heard clearly from users that great AI should not only be smart, but enjoyable to talk to.”
~Open AI.
According to OpenAI, queries will be automatically matched to the models that can answer them best.
The real kicker with this update is the new personalization features added alongside the models. OpenAI would be expanding the personality presets for the conversational tone of the models “to better reflect the most common ways people use ChatGPT.” The options now include “Professional,” “Nerdy,” “Cynical,” “Friendly,” “Candid,” “Efficient” and “Quirky.” The default setting also remains.
Another bright side of the update is that OpenAI decided to further yield to users’ requests for more granular approaches to customization. Now, the company says instructions given to GPT-5.1 “will stick across multiple turns.” That means that if you ask the model to use AP style capitalization for your headings, it will not forget the instruction two chats later and capitalize every initial.
The company is also working on including a way for the AI’s memory to become intuitive over time and an “experiment for new ways to fine-tune ChatGPT’s style directly from settings.” this would cover things like conciseness and emoji frequency.
“With more than 800 million people using ChatGPT, we’re well past the point of one-size-fits-all.”
Assuming the release is just as touted, it is a much-needed pivot after the very awkward rollout of GPT-5 in August. Months of hype only gave users a model only slightly better than GPT-4o, plus inconsistent reasoning and a mechanical tone. As a matter of fact, many early users complained that the GPT-5 performance in math, science and writing was subpar to older models. So, a day after launch, OpenAI reinstated GPT-4o as an option, with CEO Sam Altman blaming the flop on “router errors.”
GPT-5.1 keeps that routing system (now called Auto). However, OpenAI says it is smarter and more reliable this time.
Some users can start expecting the two new models this week, as the company says the rollout begins immediately for Pro, Plus, Go and Business users, with free users following soon. But we are not bidding the old GPT-5 models goodbye just yet. The company said that they will be available in the dropdown menu of ChatGPT’s legacy models for three months before they disappear. This way, users who find the upgraded models too much can opt for GPT-5 or GPT-4o, which isn’t included in the sunset period for GPT-5.
As for Enterprise and Education clients, they will get a seven-day early access toggle before GPT-5.1 becomes the new default. Developers can also access the models through the API.
GPT-5.1 is OpenAI’s answer to rivals closing in. Microsoft, OpenAI’s biggest investor, has diversified by integrating Anthropic’s Claude models into Copilot Researcher, GitHub Copilot, and its new Office Agent. Meanwhile, Baidu’s ERNIE-4.5-VL-28B-A3B-Thinking recently outperformed GPT-5 in some reasoning benchmarks.
The new models also come as OpenAI has been expanding its ecosystem. Just weeks ago, it introduced ChatGPT Atlas, an AI-powered web browser with an “agent mode” that performs actions on behalf of users.