NVIDIA shows how to link eight DGX Spark units to run frontier-class models
Announced October 22, 2026
Verified (fact)
The NVIDIA developer blog described a setup that connects eight DGX Spark nodes to run frontier-class AI models. It targets developers who want to run multiple models in parallel and chain inference calls. Specific performance figures are not in the excerpt.
Why it matters (interpretation)
It is NVIDIA's answer to demand for running large open models locally by linking desktop-class hardware.
Sources
- Official Running Frontier-Class AI Models Across 8 NVIDIA DGX Spark Nodes NVIDIA
More about NVIDIA
-
NVIDIA reportedly releases Flora, open models that auto-label actions, success and failure causes in task videos, claiming robot data costs cut up to 125x
NVIDIA has released Flora, a family of open models that analyze videos of humans and robots performing tasks and automatically annotate...
-
OpenAI, Anthropic and NVIDIA each post record Q3 lobbying spend: NVIDIA $2.2M, Anthropic $2.1M, OpenAI $1.5M (report)
According to federal disclosures, Anthropic, OpenAI and NVIDIA all recorded their highest quarterly lobbying spending in July–September,...
-
Mercedes-Benz brings MB.DRIVE ASSIST PRO, urban driving assistance co-developed with NVIDIA, to Germany (report)
Mercedes-Benz announced in Stuttgart on the 21st that it will introduce MB.DRIVE ASSIST PRO, an AI-based driver assistance system built...
-
NVIDIA expands its Cosmos world model platform with training tools, simulation checkpoints and a 'Cosmos Data Factory' blueprint
NVIDIA has added new features to Cosmos, its platform for building world models for physical AI (models that help robots and other...
-
After GTC Berlin, NVIDIA shares European open-model adoption cases in finance, retail, public sector and weather, plus 'DSX' cases targeting the power bottleneck
NVIDIA published five posts on European industry adoption cases to coincide with GTC Berlin. In finance, Revolut is rebuilding its...