Friday, Nov 13 | --:--
Back to home

Nvidia DGX Spark 64GB: $4,999 Local AI Box On Sale Oct. 23

Nvidia’s October 2 blog says a 64GB unified-memory DGX Spark configuration — same GB10 Grace Blackwell Superchip, DGX OS, and Nvidia AI stack as the higher-memory SKU — goes on sale Friday, Oct. 23 from Acer, ASUS, Dell, Gigabyte, HP, and MSI, starting at $4,999. Nvidia says it supports up to 100-billion-parameter models on device; two units can pool to 128GB. The up-to-1.7x cluster figure is Nvidia’s own Qwen 3.8 27B test.

Times of AI Desk 4 min read Santa Clara, CA View as Markdown
Engraved Nvidia mark on a compact DGX Spark chassis with a 64GB memory plate and a twin-unit Sync ribbon, security-intaglio in green and graphite.

Correction, October 14, 2026: We removed an unsupported description of the 64GB configuration as a cheaper tier. Its price relative to the 128GB model has not been verified.

Local agent hardware just got a lower-memory configuration — priced by Nvidia, timed for late October, with cluster claims that are still a single lab test.

Nvidia, in an October 2, 2026 blog by Allen Bourgoyne, says DGX Spark will be available with 64GB of unified memory starting Friday, Oct. 23, from Acer, ASUS, Dell, Gigabyte, HP, and MSI, starting at $4,999. The 64GB SKU keeps the GB10 Grace Blackwell Superchip, DGX OS, and full Nvidia AI software stack. Nvidia says it supports up to 100-billion-parameter models on device. Two units can connect via built-in ConnectX-7 networking and Nvidia Sync Cluster Assistant, pooling memory to 128GB and expanding support to up to 200 billion parameters.

Claims vs checks#

In Nvidia’s Qwen 3.8 27B test, two clustered 64GB systems delivered up to 1.7× the performance of one system. That is a single vendor test on one model and does not establish a general speedup. Street pricing relative to the 128GB SKU and independent local-inference benchmark results are not verified here.

Limits#

  • This account relies on Nvidia’s blog; independent pricing and third-party cluster performance are not verified here.
  • 1.7× is Nvidia’s own Qwen 3.8 27B cluster result, not a general speedup claim.
  • Parameter ceilings (100B / 200B) and OEM list are company-stated.
  • Independent Elo or Artificial Analysis rankings are not verified here.

Sources#

Scroll to continue reading