OpenAI Unveils ‘Ultrafast’: A Revolutionary Mode Boosting GPT-5.6 Sol’s Performance to 14 Times the Speed
For those who have wished for ChatGPT to respond faster, OpenAI is now providing a solution.
The AI research organization has introduced a new mode known as Ultrafast, which aims to significantly enhance the operational speed of its latest and advanced model, GPT-5.6 Sol.
According to the company, the Ultrafast mode operates at 14 times the standard processing speed, generating as many as 750 output tokens per second. These tokens represent the unique segments of text produced by a language model during interactions with users.
“Previously, achieving real-time efficiency often required opting for a smaller or more specialized model,” stated the company in a Thursday blog update. “Ultrafast signifies a new advancement: achieving more effective results per second.”
Other competitors in the field, such as Anthropic, have also introduced faster versions of their models. For instance, Claude offers a fast mode, though it does not match the speed that OpenAI is currently achieving.
OpenAI proposes that this accelerated version of GPT-5.6 Sol can be utilized in various corporate applications, particularly in areas like incident response, customer support, financial market analysis, and e-commerce, among others.
Currently in preview, Ultrafast is fueled by OpenAI’s collaboration with chip manufacturer Cerebras. At this stage, access to the preview is limited to a select group of customers, but OpenAI plans to broaden availability as “capacity increases.”
