- Регистрация
- 17 Февраль 2018
- Сообщения
- 39 597
- Лучшие ответы
- 0
- Реакции
- 0
- Баллы
- 2 093
Offline
New controls attempt to please critics on both sides with a balance between bland and habit-forming.
Credit: Chris Madden via Getty Images
On Wednesday, OpenAI released GPT-5.1 Instant and GPT-5.1 Thinking, two updated versions of its flagship AI models now available in ChatGPT. The company is wrapping the models in the language of anthropomorphism, claiming that they’re warmer, more conversational, and better at following instructions.
The release follows complaints earlier this year that its previous models were excessively cheerful and sycophantic, along with an opposing controversy among users over how OpenAI modified the default GPT-5 output style after several suicide lawsuits.
The company now faces intense scrutiny from lawyers and regulators that could threaten its future operations. In that kind of environment, it’s difficult to just release a new AI model, throw out a few stats, and move on like the company could even a year ago. But here are the basics: The new GPT-5.1 Instant model will serve as ChatGPT’s faster default option for most tasks, while GPT-5.1 Thinking is a simulated reasoning model that attempts to handle more complex problem-solving tasks.
OpenAI claims that both models perform better on technical benchmarks such as math and coding evaluations (including AIME 2025 and Codeforces) than GPT-5, which was released in August.
Improved benchmarks may win over some users, but the biggest change with GPT-5.1 is in its presentation. OpenAI says it heard from users that they wanted AI models to simulate different communication styles depending on the task, so the company is offering eight preset options, including Professional, Friendly, Candid, Quirky, Efficient, Cynical, and Nerdy, alongside a Default setting.
These presets alter the instructions fed into each prompt to simulate different personality styles, but the underlying model capabilities remain the same across all settings.
An illustration showing GPT-5.1’s eight personality styles in ChatGPT. Credit: OpenAI
In addition, the company trained GPT-5.1 Instant to use “adaptive reasoning,” meaning that the model decides when to spend more computational time processing a prompt before generating output.
The company plans to roll out the models gradually over the next few days, starting with paid subscribers before expanding to free users. OpenAI plans to bring both GPT-5.1 Instant and GPT-5.1 Thinking to its API later this week. GPT-5.1 Instant will appear as gpt-5.1-chat-latest, and GPT-5.1 Thinking will be released as GPT-5.1 in the API, both with adaptive reasoning enabled. The older GPT-5 models will remain available in ChatGPT under the legacy models dropdown for paid subscribers for three months.
The company says it wants to give people time to compare model outputs and adapt at their own pace and that going forward, it will communicate deprecation periods clearly with advance notice. OpenAI also published a system card with information on its safety approach for GPT-5.1.
Seeking balance
In a blog post published Wednesday, OpenAI CEO of Applications Fidji Simo wrote that the company wants ChatGPT to “feel like yours and work with you in the way that suits you best.” Simo wrote that with more than 800 million people using ChatGPT, the company has moved past one-size-fits-all approaches. She wrote that people experience ChatGPT in individual ways, with some wanting direct and neutral responses while others prefer different output patterns.
The preset “personality” options work by injecting different instructions into the system prompt that the model processes before generating each response. OpenAI says the original Cynical and Nerdy options from earlier this year will remain available in the personalization settings dropdown.
For users who want more control over outputs, OpenAI is experimenting with options to adjust specific characteristics from personalization settings, including how concise responses are and how frequently the model generates emojis. ChatGPT can also offer to update these settings during conversations when it detects users requesting certain output patterns. The company says updates to personalization settings now take effect across all chats immediately, including ongoing conversations.
A benchmark chart from OpenAI. “GPT‑5.1 Thinking varies its thinking time more dynamically than GPT‑5 Thinking,” the company writes. Credit: OpenAI
Simo addressed the balance between customization and accuracy in her blog post. “Personalization taken to an extreme wouldn’t be helpful if it only reinforces your worldview or tells you what you want to hear,” she wrote. She compared excessive customization to editing a spouse’s traits to always agree, noting that “the best people in our lives are the ones who listen and adapt, but also challenge us and help us grow.”
That concern about excessive personalization is not theoretical. Amid a year full of accusations of AI chatbots inspiring suicides and people descending into obsessive fantasy-rabbit-hole scenarios, OpenAI recently released safety research that details its plan to deal with people who develop unhealthy attachments to its AI chatbots. The company says these situations are rare, but it is working with an expert council and mental health clinicians to understand what healthy interactions with AI models should look like.
Even so, the root problem is arguably that ChatGPT still pretends to be a person—a consistent entity that knows you and learns your preferences over time. It assumes the mantle of human emotion and acts like it understands you and sympathizes with what you’re going through, which could potentially lead users into the same kind of thorny situations we’ve seen repeatedly in the past.
A screenshot example of new “warmer” GPT-5.1 outputs presented on the OpenAI website. Credit: OpenAI
It’s a tricky position for OpenAI to be in. When the company changes ChatGPT’s output style to be too reserved and robotic, it gets complaints from one set of users. When the models are too warm, the company receives criticism from experts who worry about how the models might affect vulnerable users. The new personality choices are OpenAI’s attempt to balance the needs of a broad spectrum of users who approach its chatbot with vastly different use cases, from programming assistance to being a virtual best friend.
Meanwhile, the company faces a fundamental business tension between making AI models engaging enough for widespread adoption while attempting to avoid inspiring user behavior that could become harmful. Simo addressed some of these concerns in her blog post. “We also have to be vigilant about the potential for some people to develop attachment to our models at the expense of their real world relationships, well being, or obligations,” she wrote. “There will be many new challenges as this technology evolves and people use it in new ways. Building at this scale means never assuming we have all the answers.”
Credit: Chris Madden via Getty Images
On Wednesday, OpenAI released GPT-5.1 Instant and GPT-5.1 Thinking, two updated versions of its flagship AI models now available in ChatGPT. The company is wrapping the models in the language of anthropomorphism, claiming that they’re warmer, more conversational, and better at following instructions.
The release follows complaints earlier this year that its previous models were excessively cheerful and sycophantic, along with an opposing controversy among users over how OpenAI modified the default GPT-5 output style after several suicide lawsuits.
The company now faces intense scrutiny from lawyers and regulators that could threaten its future operations. In that kind of environment, it’s difficult to just release a new AI model, throw out a few stats, and move on like the company could even a year ago. But here are the basics: The new GPT-5.1 Instant model will serve as ChatGPT’s faster default option for most tasks, while GPT-5.1 Thinking is a simulated reasoning model that attempts to handle more complex problem-solving tasks.
OpenAI claims that both models perform better on technical benchmarks such as math and coding evaluations (including AIME 2025 and Codeforces) than GPT-5, which was released in August.
Improved benchmarks may win over some users, but the biggest change with GPT-5.1 is in its presentation. OpenAI says it heard from users that they wanted AI models to simulate different communication styles depending on the task, so the company is offering eight preset options, including Professional, Friendly, Candid, Quirky, Efficient, Cynical, and Nerdy, alongside a Default setting.
These presets alter the instructions fed into each prompt to simulate different personality styles, but the underlying model capabilities remain the same across all settings.
An illustration showing GPT-5.1’s eight personality styles in ChatGPT. Credit: OpenAI
In addition, the company trained GPT-5.1 Instant to use “adaptive reasoning,” meaning that the model decides when to spend more computational time processing a prompt before generating output.
The company plans to roll out the models gradually over the next few days, starting with paid subscribers before expanding to free users. OpenAI plans to bring both GPT-5.1 Instant and GPT-5.1 Thinking to its API later this week. GPT-5.1 Instant will appear as gpt-5.1-chat-latest, and GPT-5.1 Thinking will be released as GPT-5.1 in the API, both with adaptive reasoning enabled. The older GPT-5 models will remain available in ChatGPT under the legacy models dropdown for paid subscribers for three months.
The company says it wants to give people time to compare model outputs and adapt at their own pace and that going forward, it will communicate deprecation periods clearly with advance notice. OpenAI also published a system card with information on its safety approach for GPT-5.1.
Seeking balance
In a blog post published Wednesday, OpenAI CEO of Applications Fidji Simo wrote that the company wants ChatGPT to “feel like yours and work with you in the way that suits you best.” Simo wrote that with more than 800 million people using ChatGPT, the company has moved past one-size-fits-all approaches. She wrote that people experience ChatGPT in individual ways, with some wanting direct and neutral responses while others prefer different output patterns.
The preset “personality” options work by injecting different instructions into the system prompt that the model processes before generating each response. OpenAI says the original Cynical and Nerdy options from earlier this year will remain available in the personalization settings dropdown.
For users who want more control over outputs, OpenAI is experimenting with options to adjust specific characteristics from personalization settings, including how concise responses are and how frequently the model generates emojis. ChatGPT can also offer to update these settings during conversations when it detects users requesting certain output patterns. The company says updates to personalization settings now take effect across all chats immediately, including ongoing conversations.
A benchmark chart from OpenAI. “GPT‑5.1 Thinking varies its thinking time more dynamically than GPT‑5 Thinking,” the company writes. Credit: OpenAI
Simo addressed the balance between customization and accuracy in her blog post. “Personalization taken to an extreme wouldn’t be helpful if it only reinforces your worldview or tells you what you want to hear,” she wrote. She compared excessive customization to editing a spouse’s traits to always agree, noting that “the best people in our lives are the ones who listen and adapt, but also challenge us and help us grow.”
That concern about excessive personalization is not theoretical. Amid a year full of accusations of AI chatbots inspiring suicides and people descending into obsessive fantasy-rabbit-hole scenarios, OpenAI recently released safety research that details its plan to deal with people who develop unhealthy attachments to its AI chatbots. The company says these situations are rare, but it is working with an expert council and mental health clinicians to understand what healthy interactions with AI models should look like.
Even so, the root problem is arguably that ChatGPT still pretends to be a person—a consistent entity that knows you and learns your preferences over time. It assumes the mantle of human emotion and acts like it understands you and sympathizes with what you’re going through, which could potentially lead users into the same kind of thorny situations we’ve seen repeatedly in the past.
A screenshot example of new “warmer” GPT-5.1 outputs presented on the OpenAI website. Credit: OpenAI
It’s a tricky position for OpenAI to be in. When the company changes ChatGPT’s output style to be too reserved and robotic, it gets complaints from one set of users. When the models are too warm, the company receives criticism from experts who worry about how the models might affect vulnerable users. The new personality choices are OpenAI’s attempt to balance the needs of a broad spectrum of users who approach its chatbot with vastly different use cases, from programming assistance to being a virtual best friend.
Meanwhile, the company faces a fundamental business tension between making AI models engaging enough for widespread adoption while attempting to avoid inspiring user behavior that could become harmful. Simo addressed some of these concerns in her blog post. “We also have to be vigilant about the potential for some people to develop attachment to our models at the expense of their real world relationships, well being, or obligations,” she wrote. “There will be many new challenges as this technology evolves and people use it in new ways. Building at this scale means never assuming we have all the answers.”
