OpenAI is preparing to expand access to its Ultrafast API, according to recent updates in the company's platform documentation. The feature is currently being integrated into the Responses API Playground, where users will be able to select between Standard, Fast, and Ultrafast processing speeds.
The Ultrafast mode, which features the GPT-5.6 Sol model, offers output speeds of up to 750 tokens per second. This represents a performance increase of up to 14 times compared to the Standard mode. Currently, this high-speed functionality remains restricted to a limited group of customers.
Speculation suggests that OpenAI may announce a wider rollout of the Ultrafast API during the upcoming DevDay event on September 29. The company has already introduced the GPT-6 series, including the GPT-6 Sol and GPT-6 Astra models, though it remains unconfirmed if these will support the Ultrafast mode at launch.
