Accelerate Your AI Workflows Locally with the NVIDIA DGX Spark GB10
Build, fine-tune, and deploy massive models right from your desk. The DGX Spark combines the Grace Blackwell GB10 Superchip with 128GB of unified memory to support your most demanding AI, simulation, and professional visual projects.
Designed for Professional Performance
Professionals require secure, immediate access to high-compute environments. The DGX Spark delivers this power directly to your workspace.
- Run Massive Models Locally: Develop and test reasoning models with up to 200 billion parameters directly on your workstation. Keep your intellectual property secure and always accessible.
- Seamless Shared Memory: The 128GB LPDDR5x unified memory architecture allows the 20-core ARM CPU and the Blackwell GPU to share data instantly, speeding up rendering and data orchestration.
- Exceptional Compute Output: Experience 1 PFLOP of AI performance to push your research, generative AI applications, and visual effects forward.
- Scale Your Setup: Built-in ConnectX-7 networking lets you link two DGX Spark systems together to handle models exceeding 400 billion parameters.
- Built for AI Developers & Data Scientists: Iterate faster with the pre-installed NVIDIA AI software stack, including tools like PyTorch, Jupyter, and Ollama ready out of the box.
Key Specifications
| Processor | NVIDIA Grace 20-Core ARM CPU |
| Graphics | NVIDIA Blackwell GPU |
| Memory | 128 GB LPDDR5x Unified System Memory |
| Storage | Upto 4TB NVMe M.2 SSD with self-encryption |
| Networking | 10 Gigabit Ethernet & dual NVIDIA ConnectX-7 Ports |