OpenAI and Cerebras Launch GPT-5.6 Sol Ultrafast, Delivering Frontier AI at 14× Speed
OpenAI and Cerebras unveiled Ultrafast on August 13, 2026, a new API tier running GPT-5.6 Sol at up to 750 output tokens per second, up to 14 times faster than Standard processing. The announcement follows a sweeping infrastructure commitment formalized in January 2026, under which Cerebras committed to deliver 750 megawatts of compute capacity through 2028, with the deal valued at over $10 billion. OpenAI started with a small group of companies across coding, financial research, voice AI, and e-commerce to study where the speed creates real value before expanding access. For developers, the pitch is that the traditional trade-off between model quality and response speed is now optional, at least for businesses willing to join a waitlist for a tier with no published price. On GDP-Val, a benchmark for economically valuable knowledge work, Ultrafast delivered a 5.6× end-to-end speedup with no quality degradation. The move signals OpenAI's pivot toward specialized inference hardware as the path to scale frontier intelligence deployment rather than relying solely on GPU providers.