OpenAI Previews Ultrafast API Tier Running GPT-5.6 Sol at 14X Standard Speed
The short answer
OpenAI is previewing a new API tier powered by Cerebras that delivers up to 750 tokens per second — a potential threshold-crosser for small businesses evaluating live customer-facing AI. Response speed directly affects user experience in customer-facing AI applications.
- OpenAI
- GPT-5.6
- API
- speed
- Cerebras
- customer service
What happened
OpenAI is previewing a new API service tier called Ultrafast that runs GPT-5.6 Sol at up to 14 times the speed of standard inference. Powered by chip maker Cerebras, the tier delivers up to 750 output tokens per second, making it among the fastest publicly available options for developers building commercial applications on GPT-5.6 Sol.
The preview is aimed at enterprise users and developers who need real-time AI responses. Use cases like live customer chat, automated phone response, and interactive document generation are all scenarios where current latency is a barrier to natural-feeling interactions. OpenAI announced the tier alongside a developer-focused builder's guide for GPT-5.6. Pricing has not been published; Ultrafast is currently in preview under enterprise terms.
What it means for your business
Response speed directly affects user experience in customer-facing AI applications. At 750 tokens per second, a 200-word reply generates in roughly two seconds. If you are evaluating AI-powered live chat or automated phone response, the Ultrafast tier — once pricing is public — could make those applications feel noticeably more natural. For back-office tasks like drafting emails or summarizing notes, standard inference speeds are already sufficient. Hold off on building any product around Ultrafast's speed until pricing and availability terms are confirmed.
Source: OpenAI