OpenAI announced the launch of the GPT-6 Astra Ultrafast model, now available via the OpenAI API and for eligible ChatGPT Work and Codex users. The Ultrafast mode runs on NVIDIA Blackwell GPUs and offers 8 times faster token generation speed than Astra's standard mode, achieved through inference optimizations that leverage the Blackwell architecture's capabilities within OpenAI models. OpenAI stated that the faster generation speed can shorten the edit-test-debug cycle for coding agents, reduce the time to generate responses between tool calls, and make interactive applications more responsive.