Nvidia is launching a 64GB configuration of its DGX Spark personal AI supercomputer on October 23, starting at $4,999 and shipping from Acer, ASUS, Dell, Gigabyte, HP and MSI. The new SKU keeps the GB10 Grace Blackwell Superchip and full Nvidia AI software stack of the existing 128GB model while hitting a lower entry price for developers running agents and inference workloads on device.
“Local AI is becoming more useful by the token.”— Allen Bourgoyne, Nvidia
The 64GB system supports up to 100-billion-parameter models running entirely locally, with no cloud dependency. Each unit ships with DGX OS, CUDA-X AI libraries, the Nvidia Agent Toolkit, Nemotron open models, and runtimes including Ollama, vLLM and PyTorch pre-configured.
Every DGX Spark also includes a built-in Nvidia ConnectX-7 NIC, which lets two units connect directly over a QSFP cable to pool memory to 128GB and extend model support to 200 billion parameters. On Nvidia's Qwen 3.8 27B test, two clustered 64GB systems delivered up to 1.7x the performance of a single unit over the 200 GbE fabric.
“As AI agents move from experiments into everyday development, increasingly capable open models are shrinking to fit on more devices, giving builders more to run locally.”— Allen Bourgoyne, Nvidia
Key facts
- 01DGX Spark 64GB ships October 23 starting at $4,999 from Acer, ASUS, Dell, Gigabyte, HP and MSI.
- 02The system supports up to 100-billion-parameter models on device using the GB10 Grace Blackwell Superchip.
- 03Two 64GB units clustered via ConnectX-7 pool to 128GB memory and support up to 200B-parameter models.
- 04In Nvidia's Qwen 3.8 27B test, two clustered 64GB systems hit 1.7x the performance of a single unit.
The Nvidia Sync Cluster Assistant handles the multi-node setup: it detects connected units, validates device configuration and configures the ConnectX-7 network. Because every node runs the same software stack, workloads scale from one unit to two without reconfiguration.
Nvidia also plans to release the Sync Model Launcher at the end of the month. The tool lets developers download and launch Qwen3.8 27B on a single DGX Spark or a cluster, with the model made accessible from a user's laptop and OpenCode wired up for in-browser development.
Blender is listed as one of the first major creator applications to support the platform, with a prebuilt installer coming soon. Nvidia is also pointing developers to agentic playbooks for NemoClaw, OpenClaw, Hermes Agent and OpenShell on build.nvidia.com.
The 64GB launch arrives alongside new Windows PCs powered by Nvidia RTX Spark, due this month from Acer, ASUS, Dell, HP, Lenovo, Microsoft and MSI. Nvidia will detail more at GTC Berlin, running October 20-22.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
