OpenAI's Ultrafast Mode Runs GPT-5.6 Sol 14x Faster

OpenAI has launched Ultrafast, a new mode for its GPT-5.6 Sol model that processes up to 750 output tokens per second, 14 times faster than standard processing. The company says this marks progress toward delivering more useful work per second without sacrificing model capability. Powered by a partnership with chipmaker Cerebras, Ultrafast is currently in limited preview for select customers. OpenAI says it targets corporate use cases including incident response, customer service, financial analysis, and e-commerce, with broader access planned as capacity expands.