OpenAI, 'Ultra-fast mode' up to 14x faster GPT-5.6 Sol
Auto-translated to English. Show original
On the left is Ultrafast mode, and on the right is Standard. Ultrafast mode completes generation first.
On the 14th, OpenAI began offering 'Ultrafast mode,' which runs 'GPT-5.6 Sol' up to 14 times faster than standard processing, as a limited preview for select users. It is being rolled out via the OpenAI API and generates up to 750 output tokens per second.
Ultrafast adopts Cerebras technology, a mechanism that increases processing speed while maintaining the performance of GPT-5.6 Sol. It is intended for use in products and operations that require real-time performance, and testing is underway with companies in fields such as coding, commerce, financial research, and customer support.
— OpenAI (@OpenAI)August 13, 2026Previewing Ultrafast mode: GPT-5.6 Sol at up to 14x the speed.
Launching first in the OpenAI API to a select group of customers with expanded access to more businesses as capacity grows.pic.twitter.com/a5dleofiDJ
Intended use cases include analysis of logs and code changes during system failures, evaluation of market signals and transactions, and identification of suspicious activity. In customer support, it is cited for real-time responses to inquiries involving multiple processes, and in commerce, for inventory checks, product suggestions, and resolving payment issues.
GPT-5.6 Sol in Ultrafast mode is currently available only to select users. OpenAI states that it will verify usage during the preview period and expand access as capacity increases, and is accepting registrations for the latest information on its website.
KioskNews shows a cleaned-up reading view extracted from the publisher’s page — the original always lives on their site, not ours.