NVIDIA Adds Flint Model and Management Tools to Local AI Lineup

NVIDIA outlined new software, model and enterprise management features for its local AI systems at GTC Berlin on Wednesday, adding tools for developers as its next DGX Spark configuration approaches.

RTX Spark is aimed at Windows users running personal agents, DGX Spark is designed for Linux developers, and DGX Station targets enterprise workloads. The products span individual PCs, developer systems and managed workstations. RTX Spark PCs are available, while DGX Station shipments are underway.

The next hardware step is the 64GB DGX Spark configuration, scheduled to roll out Friday, Oct. 23, through Acer, ASUS, Dell, Gigabyte, HP and MSI. The 64GB DGX Spark starts at $4,999. NVIDIA’s GTC update also detailed software, model support and management features for its broader local AI lineup.

An October DGX Spark update includes a Wi-Fi Assistant and guidance for troubleshooting and device maintenance. NVIDIA Sync’s Model Launcher, scheduled for later this month, is designed to download a compatible model and software container and connect them to OpenCode.

For DGX Station, NVIDIA listed seven new management integrations, including Microsoft Intune, Splunk by Cisco and Tanium. The system is shipping through ASUS, Dell Technologies, GIGABYTE, HP, MSI, Supermicro and Exxact. A Windows version is expected in December, with support planned for manufacturing applications and on-premises AI agents.

The model updates include Recursive’s Flint, a 40-billion-parameter mixture-of-experts model with 2.5 billion active parameters. Underdog Desktop Agent is also available for RTX Spark. These additions extend the lineup’s software and model options alongside the systems’ different roles, from personal agents to enterprise workloads.

Partner rollouts of the 64GB DGX Spark are scheduled to begin Friday, Oct. 23.