NVIDIA Launches Local AI Agents on RTX and DGX Spark

The paradigm of consumer computing has traditionally revolved around the concept of personal devices — from PCs to smartphones and tablets. Now, generative AI, particularly OpenClaw, has introduced a new category: agent computers. These devices, such as the NVIDIA DGX Spark desktop AI supercomputer or dedicated NVIDIA RTX PCs, are ideal for running personal agents — privately and for free.

This week at NVIDIA GTC, a host of new agentic AI announcements are being showcased, including new open models for local agents like NVIDIA Nemotron 3 Nano 4B and Nemotron 3 Super 120B, along with optimizations for Qwen 3.5 and Mistral Small 4. NVIDIA NemoClaw, an open-source stack for OpenClaw, optimizes OpenClaw experiences on NVIDIA devices by enhancing security and supporting local models.

Easier fine-tuning with Unsloth Studio will further improve the accuracy of open models for agentic workflows. In-person GTC attendees can swing by the NVIDIA build-a-claw event in the GTC Park, running daily through March 19, from 8 a.m. to 5 p.m. NVIDIA experts will assist guests in customizing and deploying a proactive, always-on AI assistant using their device of choice.

The next generation of local models with increasingly large context windows delivers the intelligence to run agents on PC. Combined with richer user context and powerful local tools, these advances are unlocking new possibilities on AI PCs, especially on DGX Spark, which has 128GB of unified memory supporting models with over 120 billion parameters.

Nemotron 3 Super, released last week, is a 120-billion-parameter open model with 12 billion active parameters, designed to run complex agentic AI systems. On PinchBench — a new benchmark for evaluating the performance of large language models with OpenClaw — Nemotron 3 Super scored 85.6%, making it the top open model in its class. The Mistral Small 4 model, with 119 billion parameters, is also optimized for general tasks.