Overview

• Sales begin Oct. 15 on NVIDIA.com with systems also available through OEM partners including Acer, ASUS, Dell, GIGABYTE, HP, Lenovo and MSI, plus select U.S. retail via Micro Center.
• Built on the Grace Blackwell GB10 Superchip, the workstation delivers up to 1 petaflop of AI performance with 128 GB of unified CPU‑GPU memory, NVLink‑C2C, and integrated 200 Gb/s ConnectX‑7 networking.
• NVIDIA says developers can run inference on models up to 200 billion parameters and fine‑tune models up to 70 billion parameters locally.
• Early users such as Anaconda, Cadence, ComfyUI, Docker, Google, Hugging Face, JetBrains, LM Studio, Meta, Microsoft, Ollama and Roboflow are testing and optimizing tools, with NYU’s Global Frontier Lab citing peta‑scale desktop research benefits.
• Independent reports note a Linux‑based DGX OS image rather than Windows and a starting price in the low‑$3,000 range.