OpenAI Previews Ultrafast Mode for GPT-5.6 Sol: 750 Tokens per Second via Cerebras

OpenAI previewed Ultrafast, a new API tier for GPT-5.6 Sol delivering up to 750 output tokens per second — 14x standard speed. The tier runs on Cerebras wafer-scale inference hardware, not OpenAI's own infrastructure. Access is limited to a small group of customers, with expansion planned as capacity grows. Pricing and general availability remain undisclosed.