OpenAI has introduced a new mode called Ultrafast, designed to dramatically accelerate the performance of its most advanced model, GPT-5.6 Sol. This update responds to user demand for quicker AI interactions and marks a significant milestone in optimizing both speed and capability.
What is Ultrafast?
Ultrafast enables GPT-5.6 Sol to operate at up to 14x the speed of standard processing, delivering as many as 750 output tokens per second. Tokens are the discrete units of text generated by large language models during communication, making this speed leap particularly impactful for real-time applications.
"Until now, getting real-time speed typically meant choosing a smaller or more specialized model," OpenAI noted in a Thursday blog post. "Ultrafast points to progress in a new direction: more useful work per second."
Competitive Context and Applications
OpenAI is not alone in pursuing faster AI responses. Rivals like Anthropic have introduced fast mode for Claude, though that feature does not yet match the output velocity claimed by OpenAI. Ultrafast aims to bridge the gap between high intelligence and immediacy, potentially transforming several enterprise use cases:
- Incident response – faster diagnostics and resolution
- Customer service – real-time support without latency
- Financial market analysis – accelerated data interpretation
- E-commerce – instant, dynamic user interactions
Availability and Underlying Technology
Ultrafast is currently in preview, powered by OpenAI's partnership with chipmaker Cerebras. The feature is initially limited to a small group of customers, with OpenAI stating that access will expand "as capacity grows." This phased rollout underscores the company's cautious approach to maintaining performance quality while scaling infrastructure.
As AI adoption accelerates into 2026, Ultrafast could set a new benchmark for real-time language model interactions, prompting further innovation across the industry.
via TechCrunch AI