- Published on
OpenAI launches Ultrafast mode, GPT-5.6 Sol AI 14x faster
- Authors
- Name
- DeepHorus
On August 14, OpenAI announced the preview release of Ultrafast mode for its most powerful AI model, GPT-5.6 Sol. This mode can output up to 750 tokens per second, 14 times the standard processing speed, designed to meet business scenarios with extremely high real-time requirements. Currently, this mode is only available to a small group of customers to test its practical value in different workflows.
For a long time, the speed and capability of AI models have often been difficult to achieve simultaneously. OpenAI stated that users seeking faster output typically have to sacrifice model capability and instead choose smaller or more specialized models. Ultrafast mode attempts to break this trade-off, making "accomplishing more valuable work per second" a new direction. This marks an important breakthrough in the response speed of AI services.
It is reported that Ultrafast mode is technically supported by chip company Cerebras. OpenAI has not disclosed specific technical details, but emphasized that this mode aims to further improve the model's response speed in user workflows. Through this collaboration, OpenAI hopes to combine the intelligence of top-tier models with near real-time output capabilities.
The early application scenarios listed by OpenAI include incident response and reliability analysis, financial research and security analysis, customer service and voice interaction, product Q&A and inventory queries in e-commerce scenarios, as well as real-time research and experiments. These scenarios are all latency-sensitive, and speed improvements may bring significant efficiency gains. For example, in customer service scenarios, faster responses can improve user experience; in financial analysis, it helps to capture market changes in a timely manner.
Currently, this mode is in a limited preview phase and is only open to a small group of customers. OpenAI stated that these early tests help determine where speed improvements provide the most value, and also help observe whether the model can keep up with the pace of user operations. By collecting feedback, the company will further optimize the mode's performance.
The launch of Ultrafast mode reflects the AI industry's continuous pursuit of real-time performance. As model capabilities increase, speed is becoming a key dimension of competition. If this mode performs well in preview, it is expected to be promoted to a wider user base in the future. At the same time, it may also drive the implementation of more application scenarios, from enterprise-level tools to consumer services, all of which could benefit from faster AI responses.
Industry analysts believe that OpenAI's collaboration with Cerebras demonstrates its open strategy in the hardware ecosystem. By leveraging the advantages of specialized chips, OpenAI can improve speed without sacrificing model capability. Whether this mode becomes the norm will depend on the results of preview testing and actual market demand. IT Home will continue to follow subsequent developments.
