Previewing Ultrafast mode: GPT-5.6 Sol at up to 14X the speed
Ultrafast mode is available in limited preview via the OpenAI API, powered by Cerebras, at up to 750 output tokens per second.
- Early customers testing Ultrafast include Jane Street, Podium, Basis, and Rogo.
- OpenAI uses Ultrafast internally for incident response and to condense overnight research runs into same-day iterations.
- Access expands as capacity grows; businesses can sign up for updates.
For developers and businesses using the OpenAI API, Ultrafast removes the speed-intelligence trade-off, enabling real-time workflows with frontier intelligence.
Sources
Read this as text
Back to the AI news