NVIDIA DGX Spark™ – Personal AI Desktop Supercomputer
NVIDIA DGX Spark™ – Personal AI Desktop Supercomputer – Desktop GB10 Grace Blackwell Chip
- Supercomputer performance directly to your desk in a compact, energy-efficient design, enabling enterprise-scale AI and high-performance computing right where you need it.
- The power of Grace Blackwell architecture, delivering up to 1 petaFLOP of AI performance for local model fine-tuning, inference, and analytics, accelerating your time-to-solution.
- Designed from the ground up to build and run AI, delivering seamless integration of the full NVIDIA AI software stack —so you can develop locally and deploy anywhere.
- NVIDIA DGX Spark gives you the freedom to experiment, prototype, and innovate faster by augmenting laptop, desktop, cloud, or data center resources. With more power to learn, prototype, test, and innovate, NVIDIA DGX Spark delivers exceptional ROI for increased productivity.
- Use NVIDIA DGX Spark to unlock new ideas and experiment with large models (up to 200 billion parameters at FP4) directly on your desktop with 128GB of unified memory. Empower rapid testing, validation, and iteration—driving innovation in a secure, high-performance setting.
NVIDIA DGX Spark
The video showcases the product in use.The video guides you through product setup.The video compares multiple products.The video shows the product being unpacked.NVIDIA DGX Spark
Personal AI Desktop Supercomputer
Supercomputer performance directly to your desk in a compact, energy-efficient design, enabling enterprise-scale AI and high-performance computing right where you need it.
Grace Blackwell Architecture
The power of Grace Blackwell architecture, delivering up to 1 petaFLOP of AI performance for local model fine-tuning, inference, and analytics, accelerating your time-to-solution.
NVIDIA DGX Spark Technical Specifications
| Architecture | NVIDIA Grace Blackwell |
| GPU | NVIDIA Blackwell Architecture |
| CPU | 20 core Arm, 10 Cortex-X925 + 10 Cortex-A725 Arm |
| CUDA Cores | NVIDIA Blackwell Generation |
| System Memory | 128 GB LPDDR5x, coherent unified system memory |
| Memory Interface | 256-bit |
| Memory Bandwidth | Up to 273 GB/s |
| Storage | 4TB NVME.M2 with self-encryption |
| System Dimensions | 150 mm L x 150 mm W x 50.5 mm H |
Develop, test, and validate
NVIDIA DGX Spark provides a platform for developers to create, test, and validate AI models
Fine-tune AI models up to 70 billion parameters
Improve the performance of pre-trained models by fine-tuning on NVIDIA DGX Spark
High-performance data science at your desk
NVIDIA DGX Spark delivers 128GB unified memory and 1 petaFLOP of AI performance for complex AI tasks
Test AI models up to 200 billion parameters
Fifth-generation Tensor Cores accelerate inference of AI models to test & deploy from the DGX Spark
Develop edge applications with NVIDIA AI framework
NVIDIA frameworks include Isaac, Metropolis, and Holoscan
| SKU: | B0FWJ16CCH |
| Dimensions: | 9.5 x 9.5 x 6 inches |
| Brand: | NVIDIA |
| Model: | DGX Spark |
| Colour: | Gold |
| Manufacture: | NVIDIA |
| Colour: | Gold |
Product description
NVIDIA DGX Spark
The video showcases the product in use.The video guides you through product setup.The video compares multiple products.The video shows the product being unpacked.NVIDIA DGX Spark
Personal AI Desktop Supercomputer
Supercomputer performance directly to your desk in a compact, energy-efficient design, enabling enterprise-scale AI and high-performance computing right where you need it.
Grace Blackwell Architecture
The power of Grace Blackwell architecture, delivering up to 1 petaFLOP of AI performance for local model fine-tuning, inference, and analytics, accelerating your time-to-solution.
NVIDIA DGX Spark Technical Specifications
Architecture NVIDIA Grace Blackwell GPU NVIDIA Blackwell Architecture CPU 20 core Arm, 10 Cortex-X925 + 10 Cortex-A725 Arm CUDA Cores NVIDIA Blackwell Generation System Memory 128 GB LPDDR5x, coherent unified system memory Memory Interface 256-bit Memory Bandwidth Up to 273 GB/s Storage 4TB NVME.M2 with self-encryption System Dimensions 150 mm L x 150 mm W x 50.5 mm H Develop, test, and validate
NVIDIA DGX Spark provides a platform for developers to create, test, and validate AI models
Fine-tune AI models up to 70 billion parameters
Improve the performance of pre-trained models by fine-tuning on NVIDIA DGX Spark
High-performance data science at your desk
NVIDIA DGX Spark delivers 128GB unified memory and 1 petaFLOP of AI performance for complex AI tasks
Test AI models up to 200 billion parameters
Fifth-generation Tensor Cores accelerate inference of AI models to test & deploy from the DGX Spark
Develop edge applications with NVIDIA AI framework
NVIDIA frameworks include Isaac, Metropolis, and Holoscan
I am pretty knowledgeable with Linux terminal/command prompts which makes it very user friendly for my skill set. I usually SSH in from any of my other computers. For what I don’t know, I use different LLM for helping solve the task that I work on.
I am currently using at as a hobby dev project on my home lab. Its definitely been a great asset with helping make other agents for other task.
I have very large LLM running locally. One of them is helping with discover more about my business and better ways I can run it.
Its very quiet very fast.
If you are considering ordering one, if you don’t have another computer you can login via phone to setup your connection to wifi. It doesn’t come with usb-c keyboard or mouse. If you have old USB keyboard and mouse, you will need USB to usb-c converter, if you don’t know how to ssh from another computer to the DGX Spark
Purchased just for OpenClaw.
Its doing amazing at running all free and uncensored models through ollama, OpenClaw has full admin rights of the system and downloads whatever it needs, uses ComfyUI to generate images and products.
Huge amount of resources and it loads massive models with zero problems, fast responses the same as the cloud services or quicker.
I’m runninng the qwen 3.6:27B model through ollama and opencode. Reviewing codebases and finding issues, mapping to schematics and tracing runtime problems. This system is letting me use current tools on an ITAR codebase without having to worry about code exposure and yes I am getting results in an accepatable time. No its not as fast as running with Gemini or Claude but it is entirely local and secure.
Excellent for local LLM research.
It gave me a shock when it appeared not to boot up after first setting it up. It turned out it just needed some time and started up after a while.
That was not at all obvious though. It’s basically silent and it doesn’t have any lights to show that it’s even on. I feel like nvidia could at least have forked out for an off/on light on the case.