NVIDIA DGX Spark 64GB Launches for Local AI Development

NVIDIA is expanding its local AI hardware lineup with a new DGX Spark configuration featuring 64GB of unified memory. Available this month from manufacturer partners including Acer, ASUS, Dell, Gigabyte, HP, and MSI, the new SKU starts at $4,999. This release provides developers with a more accessible entry point to the Grace Blackwell architecture while retaining the full NVIDIA AI software stack and DGX OS.

Performance and Scaling

The 64GB model supports up to 100-billion-parameter models and agentic applications entirely on-device. For workloads requiring more capacity, two units can be clustered using the NVIDIA Sync Cluster Assistant. In NVIDIA’s Qwen 3.8 27B test, two clustered 64GB systems delivered up to 1.7x performance compared to a single system. Connecting two units via QSFP cable pools their memory to 128GB, expanding model support to up to 200 billion parameters and doubling memory bandwidth.

Software and Ecosystem

The platform ships ready for agent development with support for NVIDIA Agent Toolkit, CUDA-X AI libraries, Nemotron open models, and popular runtimes like Ollama, vLLM, and PyTorch. Developers can go from power-on to running models in minutes. Blender is among the first major creator applications to support the platform, with a prebuilt installer coming soon. Additionally, the NVIDIA Sync Model Launcher will allow users to easily download and launch models like Qwen3.8 27B across connected devices.