NVIDIA AI PC said on April 2, 2026 that the new Gemma 4 models are optimized for RTX GPUs and DGX Spark, with the 26B and 31B variants aimed at local agentic AI. NVIDIA's official blog says the collaboration spans RTX PCs, workstations, DGX Spark, Jetson Orin Nano, and data center deployments, with native tool use, multimodal inputs, and local runtime support through Ollama and llama.cpp.
A DGX Spark owner on LocalLLaMA argues that NVFP4 remains far from production-ready, prompting a broader debate about whether NVIDIA's premium local AI box still justifies its price.
A detailed LocalLLaMA post compared a $10K Mac Studio M3 Ultra 512GB with a similarly priced dual DGX Spark setup for running Qwen3.5 397B A17B locally. The Mac delivered 30 to 40 tok/s and easier setup, while the dual Spark build offered faster prefill and embedding performance at much higher operational complexity.
At CES on January 5, 2026, NVIDIA introduced new AI foundation models under Cosmos and launched DGX Spark and DGX Station. The announcement links synthetic world modeling for robotics with desktop-to-datacenter AI compute.
In its February 12, 2026 post, NVIDIA describes DGX Spark as a desktop AI system now used across universities for on-prem model development and rapid iteration. The examples span South Pole neutrino analysis, medical report evaluation, and campus robotics workloads.