OpenAI to Expand Ultrafast API Access and Add a Speed Selector to Playground
OpenAI is preparing to widen access to Ultrafast, its Cerebras-powered API tier that runs GPT-5.6 Sol up to 14 times faster than Standard, and its Playground code now holds a hidden Standard, Fast and Ultrafast speed selector. We checked the code and the API specification ourselves. This article explains how the tiers fit together, what 750 tokens per second means, what is known about price and model support, where the speed pays off and how to prepare before DevDay.
