What happened?
OpenAI has launched its new model, GPT-6 Astra Ultrafast. It runs on NVIDIA Blackwell GPUs and is now available in the OpenAI API and to eligible ChatGPT Work and Codex users.
The speed angle
According to NVIDIA, Ultrafast benefits from inference optimizations in OpenAI's models and taps into the capabilities of the Blackwell architecture. As a result, it offers up to 8x faster token generation than Astra Standard mode.
For developers
Because the model is accessible through the API, developers can integrate Ultrafast directly into their own applications. NVIDIA's announcement notes that further details for the developer side will be shared.
Why it matters
This jump in token generation speed shortens response times and directly affects the experience in heavy-use scenarios such as chat and code generation. The model's reliance on NVIDIA's new-generation Blackwell hardware shows that hardware-software collaboration is central to AI acceleration.
