NVIDIA is opening a lower-cost door into local AI with the NVIDIA DGX Spark 64GB, a new configuration of its compact personal AI supercomputer. It keeps the same GB10 Grace Blackwell Superchip and NVIDIA AI software stack as the existing 128GB version, but ships with less unified memory at a more accessible price.
The 64GB DGX Spark goes on sale Friday, October 23 through Acer, ASUS, Dell, Gigabyte, HP and MSI, starting at US$4,999. NVIDIA is pitching it at developers, researchers and AI enthusiasts who want to keep agents and data on device instead of renting cloud compute for every task.
Inside the NVIDIA DGX Spark 64GB
DGX Spark packs Grace Blackwell compute, unified memory, ConnectX-7 networking and a CUDA-accelerated software stack into a chassis small enough for a desk. According to NVIDIA, the 64GB model runs models with up to 100 billion parameters entirely on device, which covers most open coding and research agents in use today.
The system ships with DGX OS, the NVIDIA Agent Toolkit, CUDA-X AI libraries and Nemotron open models. Runtimes including Ollama, vLLM and PyTorch with CUDA work from the first boot, so developers can move from power-on to running a model in minutes rather than configuring drivers and containers first.
Pairing two units with Sync Cluster Assistant
NVIDIA also announced NVIDIA Sync Cluster Assistant, a feature that connects two DGX Spark systems over a 200GbE fabric and pools their memory. Two 64GB units join into 128GB of combined memory, which NVIDIA says can support models up to 200 billion parameters. In the company’s testing with Qwen 3.8 27B, a two-system cluster delivered up to 1.7 times the performance of a single unit.
The Sync app detects the connected systems, validates their configuration and sets up the network, so the software environment stays the same when a developer moves from one unit to two. A separate NVIDIA Sync Model Launcher is set to follow at the end of October. It downloads and launches Qwen 3.8 27B on a single DGX Spark or a cluster, makes the model reachable from a laptop, and configures OpenCode to use it.
What the 64GB model changes for buyers
The new configuration slots in below the 128GB DGX Spark that NVIDIA already sells, giving smaller teams and individual developers a lower entry point into the same platform and software stack. NVIDIA first showed the concept at CES 2025 as Project DIGITS before renaming it DGX Spark.
Philippine pricing and availability for the 64GB model have not been announced. The US$4,999 starting price is the US figure, and local buyers should watch NVIDIA’s retail and distribution partners for a Philippine release and SRP.
Giancarlo Viterbo
Giancarlo Viterbo is a Filipino Technology Journalist, blogger and Editor of gadgetpilipinas.net, He is also a Geek, Dad and a Husband. He knows a lot about washing the dishes, doing some errands and following instructions from his boss on his day job. Follow him on twitter: @gianviterbo and @gadgetpilipinas.
View all articles →
