Cerebras is powering OpenAI's new Ultrafast mode for GPT-5.6 Sol, delivering 14x faster inference at up to 750 output tokens per second on the Cerebras hardware platform.
OpenAI is launching a preview of its new "Ultrafast" mode, which delivers up to 750 output tokens per second from its flagship model, GPT-5.6 Sol. The inference acceleration comes from Cerebras, which signed a ten-billion-dollar partnership with OpenAI earlier this year. Initially available only through the OpenAI API for GPT-5.6 Sol and limited to select customers, the service will expand gradually as capacity grows. Companies interested in early access can sign up through a form.
OpenAI says Ultrafast is designed to combine the speed of smaller models with the full capabilities of a large reasoning model, enabling what the company calls "more useful work per second." During incident response, for example, engineers could have logs, code changes, and reports analyzed while an outage is still happening, helping them pinpoint the cause and prepare a fix in real time. OpenAI says it's already using the model internally for this purpose.
OpenAI highlights several other scenarios. In finance, the model could evaluate market signals and flag suspicious transactions while conditions are still shifting. In customer support, complex inquiries could be resolved in real time, even when finding the answer requires multiple steps or systems. In e-commerce, it could answer product questions, check inventory levels, and personalize recommendations before a hesitant buyer abandons their cart.
Research is another area OpenAI highlights. Experiments that previously ran overnight as batch jobs could turn into interactive work sessions, letting teams test an idea, review results, adjust their approach, and kick off another run without breaking their workflow.
OpenAI already monetizes inference speed in tiers. Through the API, it offers a "Fast Mode" that promises up to 2.5x speed with lower latency for GPT-5.6 Sol at roughly double the price. Ultrafast adds a third, faster—and likely pricier—tier. The logic mirrors cloud providers like AWS, which have long charged more for the same service at higher performance levels. If speed becomes a bottleneck across industries, this tiered model gives OpenAI a direct cut of the revenue gains that faster inference creates.