OpenAI is pushing speed to the center of its latest model strategy with Ultrafast, a new mode for GPT 5.6 Sol that is being released in preview.
The company says the mode is built to accelerate how quickly its latest and most powerful model completes work, with performance reaching 14x the speed of standard processing and up to 750 output tokens per second.
What Ultrafast changes
Ultrafast is designed around a simple problem: many people want ChatGPT and related AI systems to respond more quickly, especially when the task is time-sensitive or part of a larger workflow.
According to OpenAI, the new mode gives GPT 5.6 Sol a much faster path for producing output. In practical terms, that means the model can generate far more text in less time than standard processing.
OpenAI describes output tokens as the distinct pieces of text generated by an LLM when it interacts with a human. The company says Ultrafast can deliver up to 750 output tokens per second, which is the headline measure behind its speed claim.
The shift is not only about making a chatbot feel snappier. OpenAI frames Ultrafast as a way to get more work done per second from a powerful model, rather than moving users toward smaller or more specialized models when speed matters.
"Until now, getting real-time speed typically meant choosing a smaller or more specialized model," the company said in a blog post on Thursday. "Ultrafast points to progress in a new direction: more useful work per second."
Why speed matters for GPT 5.6 Sol
GPT 5.6 Sol is described by OpenAI as its latest and most powerful model. That makes the speed of Ultrafast notable because faster AI modes often raise a tradeoff: users want rapid responses, but they may also want the capabilities of a stronger model.
OpenAI's message is that Ultrafast is meant to narrow that gap. Instead of treating real-time speed as something that requires a smaller or more specialized model, the company is presenting the new mode as a faster way to use GPT 5.6 Sol itself.
That distinction matters for businesses that use AI inside operational systems. A model that responds slowly can become a bottleneck when users need analysis, drafting, summaries, or support output in the flow of work.
By emphasizing "more useful work per second," OpenAI is positioning Ultrafast as a productivity feature as much as a technical benchmark. The point is not just that text appears faster, but that the model can support situations where delay reduces usefulness.
Where OpenAI expects companies to use it
OpenAI suggests that Ultrafast could be deployed across several corporate workflows. The examples named by the company include incident response, customer service and support, financial market analysis, and e-commerce.
Those use cases share a common need for quick output. In incident response, teams may need rapid summaries or next-step assistance. In customer service and support, faster responses can help keep conversations moving. In financial market analysis and e-commerce, the value of AI output may depend on speed as well as relevance.
The company also leaves room for other relevant areas, suggesting that Ultrafast is not limited to one narrow category of work. Its broader appeal is tied to any setting where GPT 5.6 Sol's capabilities are useful but standard processing may not be fast enough for the job.
- Incident response: faster AI output for urgent operational work.
- Customer service and support: quicker responses in user-facing workflows.
- Financial market analysis: faster generation where timing may matter.
- E-commerce: accelerated assistance for business and shopping-related tasks.
How the preview is being rolled out
Ultrafast is currently being released in preview. OpenAI says the preview is available only to a small group of customers for now.
The mode is powered by OpenAI's partnership with chipmaker Cerebras. The source article does not give additional technical details about how that partnership supports the preview, but it identifies Cerebras as the company behind the chipmaking role.
OpenAI says access will expand as "capacity grows." That means broader availability is planned, but the current rollout is limited.
The preview status is important because it signals that Ultrafast is not yet generally available to everyone. For now, the feature appears to be aimed at selected customers while OpenAI scales the underlying capacity needed to support it.
How it compares with rival fast modes
OpenAI is not alone in trying to make advanced AI models faster. The source article notes that competitors, including Anthropic, have launched accelerated versions of their models.
Claude has fast mode, but the source article says it does not deliver the kind of speed OpenAI is offering here. That comparison places Ultrafast inside a broader race to make AI systems more responsive without pushing users away from high-end models.
For OpenAI, the central claim is clear: Ultrafast makes GPT 5.6 Sol work at 14x the speed of standard processing. If the preview expands as planned, the mode could become an important option for businesses that need the strength of a leading model with response times closer to real-time work.