OpenAI has introduced a new mode called Ultrafast for its GPT-5.6 Sol model, which the company claims can operate at 14 times the speed of standard processing.
Key facts
- •Ultrafast mode is designed for OpenAI's GPT-5.6 Sol model.
- •The new mode can generate up to 750 output tokens per second.
- •OpenAI is collaborating with chipmaker Cerebras to power the feature.
- •The mode is currently in a limited preview release.
- •Targeted use cases include financial market analysis, customer support, and incident response.
OpenAI has released a new mode for its GPT-5.6 Sol model named Ultrafast, designed to significantly increase the speed at which the AI performs tasks. The company announced the update on Thursday, stating that the mode is currently available in preview to a limited group of customers.
By the numbers
Performance and Technical Capabilities
According to OpenAI, Ultrafast allows the GPT-5.6 Sol model to operate at 14 times the speed of standard processing. This capability enables the model to deliver up to 750 output tokens per second. The company noted that while real-time speed previously required using smaller or specialized models, this new mode aims to increase the amount of useful work performed per second.
Deployment and Availability
The Ultrafast mode is powered by a partnership between OpenAI and the chipmaker Cerebras. OpenAI suggests the high-speed version is suitable for corporate workflows such as customer service, incident response, financial market analysis, and e-commerce. Access to the preview is currently limited, though OpenAI stated it plans to expand availability as capacity increases.
Advertisement
This article was independently rewritten by ManyPress editorial AI from reporting originally published by TechCrunch.
