OpenAI introduced GPT-5.1 in ChatGPT on Wednesday.
The update comes four months after the generative AI vendor released GPT-5, and a day after watchdog group Public Citizen asked the vendor to remove its Sora 2 app from the market because of the multi-modal model app’s susceptibility to being used to create harmful deepfakes.
The model upgrade comes with GPT-5.1 Instant, which OpenAI said is warmer and conversational than GPT-5. The 5.1 version also boasts improved instruction following, enabling it to answer questions more reliably. The Instant feature can use adaptive reasoning to decide when to think before responding, OpenAI said.
GPT-5.1 Thinking, another component of 5.1, can now adapt the model’s thinking tasks, meaning it spends less time on easy prompts and functions and more time on complex tasks, the vendor said.
OpenAI plans to roll out GPT-5.1 gradually over the next few weeks, with an upgrade to GPT-5.1 Pro for GPT-5 Pro users soon. Meanwhile, GPT-5.1 Auto within ChatGPT will continue to route prompts or questions to the best OpenAI model that can answer them.
The vendor also renamed its customization options in ChatGPT to Default, Friendly and Efficient from Listener and Robot. It also added Professional, Candid and Quirky. Users can now tune ChatGPT’s characteristics with personalization settings, such as how concise, warm, or scannable they want its responses to be.
A trend
The update demonstrates how OpenAI is striving to create a smoother user experience for its AI chatbot, according to Arun Chandrasekaran, an analyst at Gartner.
“They want these models to be more conversational, more instruction following to the point where they hope that the entire experience of selecting the right model is completely automated,” Chandrasekaran said. “They want to automate the model selection within ChatGPT. That’s fundamentally OpenAI’s goal.”
He added that as long as OpenAI makes the model selection automation process smooth for users, then there shouldn’t be any significant problems with it. However, this update is also an example of how OpenAI is seeking room to improve its models after the original GPT-5 got mixed reviews.
“This is also a signal to the market around the pace of innovation within the company,” Chandrasekaran said.
