OpenAI preparing to expand Ultrafast API access
OpenAI appears to be preparing to broaden access to its Ultrafast API. Documentation changes suggest new speed options are coming to the platform.
AI company OpenAI seems to be preparing to expand the availability of its Ultrafast API. Recent indications from platform documentation point towards the company increasing access to its faster API services.
OpenAI is also reportedly working on new speed options for its Responses API Playground. Users may soon be able to choose between three processing speeds: Standard, Fast, and Ultrafast. This feature is currently hidden but its emergence suggests further development.
The company had previously previewed the Ultrafast mode, powered by the GPT-5.6 Sol model, stating it could achieve output speeds of up to 750 tokens per second. This represents a potential speed increase of up to 14 times compared to the Standard mode. Access to this capability has so far been limited to select clients.
An expanded rollout of the Ultrafast service may occur soon, possibly coinciding with the DevDay event scheduled for September 29. OpenAI has also introduced its GPT-6 series models, including GPT-6 Sol and GPT-6 Astra. It remains unclear whether these new models will support the Ultrafast feature upon release.