OpenAI introduces Ultrafast, the mode that accelerates GPT-5.6 Sol by up to 14 times

OpenAI took a strong step to attract corporate clients by introducing Ultrafast, a new mode designed for its most powerful model, GPT-5.6 Sol, to operate at much higher speeds. According to the company, this option allows for speeds of up to 14 times the standard processing speed and can generate up to 750 output tokens per second.

The proposal aims to address a common complaint among users of such systems: the delay in responses. Until now, achieving speeds close to real-time often meant sacrificing power and opting for smaller or specialized models. Ultrafast seeks to change that logic and demonstrate that it is possible to deliver more useful work per second without sacrificing capacity.

In the statement released this Thursday, the company emphasized this idea: “Until now, achieving real-time speed usually meant choosing a smaller or more specialized model. Ultrafast aims for progress in a new direction: more useful work per second.”

The mode is specifically designed for business workflows. Among the uses that OpenAI highlights are incident response, customer service and support, financial market analysis, and e-commerce, although the list of potential applications is broader.

Alliance with Cerebras and restricted access

Ultrafast relies on OpenAI's partnership with chip manufacturer Cerebras. This technical collaboration is what enables the announced speed levels. For now, the feature is being launched in a preview version and is only available to a small group of clients.

The company anticipates that it will expand access as available capacity grows. This is not yet a mass launch, but rather a controlled testing phase aimed at gathering experience in real high-performance environments.

Other players in the sector are also working on accelerated versions of their models. Anthropic, for example, offers a fast mode in Claude. However, OpenAI claims that the speed achieved by Ultrafast surpasses what is currently seen in those alternatives.

The bet has a clear commercial sense. By targeting companies that need almost instant responses in critical or high-volume situations, OpenAI seeks to differentiate itself and solidify its position in the corporate segment, where latency can make a concrete difference in business outcomes.