OpenAI unveils Ultrafast mode for GPT-5.6 Sol, claiming 14x speed boost

BitcoinWorld OpenAI unveils Ultrafast mode for GPT-5.6 Sol, claiming 14x speed boost OpenAI has introduced a new processing mode called Ultrafast that it says can run its latest flagship model, GPT-5.6 Sol, at up to 14 times the speed of standard inference, delivering as many as 750 output tokens per second.

The company announced the preview on Thursday, August 13, 2026, positioning it as a significant leap in real-time AI responsiveness for enterprise applications. What is Ultrafast and how does it work?

Ultrafast is a new inference mode designed specifically for GPT-5.6 Sol, OpenAI’s most advanced model to date. According to the company, the mode leverages a partnership with chipmaker Cerebras to achieve the dramatic speed increase.

In a blog post, OpenAI explained that until now, achieving real-time speed typically required selecting a smaller or more specialized model, but Ultrafast represents a new direction: more useful work per second.