GTC Spotlights NVIDIA RTX PCs and DGX Sparks Running Latest Open Models and AI Agents Locally

2026-03-17 · Source: NVIDIA Blog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Robotics & Autonomous Systems · Depth: Intermediate, medium

Summary

NVIDIA is advancing the "agent computer" paradigm, enabling generative AI agents to run privately and freely on personal devices like NVIDIA DGX Spark and RTX PCs. Key announcements from NVIDIA GTC include new open models such as Nemotron 3 Nano 4B, Nemotron 3 Super 120B, and optimizations for Qwen 3.5 and Mistral Small 4, designed for local agentic AI. NVIDIA also introduced NemoClaw, an open-source stack optimizing OpenClaw experiences on NVIDIA hardware for enhanced security and local model support. Additionally, Unsloth Studio was launched, providing a web-based UI to simplify fine-tuning of over 500 AI models, offering up to 2x faster training and 70% VRAM savings. Other updates include LTX 2.3 and FLUX.2 Klein 9B optimizations, an RTX AI video generation guide, NVIDIA AI for Media SDK updates, and the upcoming DLSS 5.

Key takeaway

For NLP Engineers and CTOs evaluating local AI agent deployment, NVIDIA's new open models and software stack, NemoClaw, offer a compelling path. You can now run powerful agents like Nemotron 3 Super 120B locally on DGX Spark or RTX PCs, significantly improving privacy and eliminating token costs. Consider integrating Unsloth Studio to efficiently fine-tune these models for your specific enterprise data and workflows, enhancing accuracy and performance.

Key insights

Generative AI agents are shifting from cloud to local devices, driven by new open models and optimized software stacks.

Principles

• Local inference enhances privacy and reduces token costs.
• Fine-tuning improves open model accuracy for specific use cases.

Method

NVIDIA NemoClaw provides an open-source stack for OpenClaw, utilizing Nemotron local models for inference and OpenShell for secure execution on NVIDIA devices.

In practice

• Use Nemotron 3 Super 120B for complex agents on DGX Spark.
• Employ Unsloth Studio for simplified fine-tuning of open models.
• Leverage NemoClaw for secure, local OpenClaw agent deployment.

Topics

• Agentic AI
• Large Language Models
• Model Fine-tuning
• NVIDIA GPUs
• Open-Source AI

Code references

• unslothai/unsloth-studio

Best for: NLP Engineer, CTO, VP of Engineering/Data, AI Engineer, Machine Learning Engineer, Software Engineer

Related on AIssential

• 😺 🎙️ Watch: NVIDIA's 120B Model Runs Like a 12B. Here's How. — NVIDIA's Nemotron-3 Super enables large AI models to run efficiently on consumer GPUs via a mixture of experts.
• From RTX to Spark: NVIDIA Accelerates Gemma 4 for Local Agentic AI — Gemma 4 models, optimized for NVIDIA GPUs, enable efficient on-device AI with multimodal and multilingual capabilities.
• NVIDIA Levels Up Local AI Agents Across RTX PCs and DGX Spark — NVIDIA is driving secure, high-performance local AI agent adoption across its RTX and DGX hardware and software platforms.
• Build a More Secure, Always-On Local AI Agent with OpenClaw and NVIDIA NemoClaw — NVIDIA NemoClaw enables secure, self-hosted AI agent deployment with local inference and robust sandboxing.
• not much happened today — The AI ecosystem is rapidly advancing open-weight multimodal agents and specialized hardware for local, efficient inference.
• Run Local AI Agents with Faster Models and Multi-Node Clustering on NVIDIA DGX Spark — Integrated hardware and software solutions significantly accelerate local AI agent development and deployment, enhancing security and scalability.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by NVIDIA Blog.