OpenAI introduced the preview Ultrafast mode—a new API pricing tier for GPT-5.6 Sol. According to OpenAI News, the model is expected to operate up to 14imes faster.
The service operates with the participation of Cerebras and, according to the published description, delivers speeds of up to 750 output tokens per second. The source does not specify access conditions or comparison parameters.
The practical significance of this launch is currently difficult to assess: the available package contains only metadata and a brief description from OpenAI News, without independent confirmation or operational details.
editorial commentary
Why it matters
If the claimed metrics are confirmed, the next observable signals will be access conditions and independent performance measurements. Potential implications for developers depend on price, limitations, and real speed stability; these data are not present in the source.