NVIDIA RTX Spark™ brings powerful AI performance on device, giving creators greater control over capacity and costs, enhanced privacy for data and context, and the flexibility to customize AI models and workflows around the way they work.
Local AI without token limits
Run and retry AI agents locally without token limits, turning variable cloud usage into controllable computing capacity while reserving the cloud for larger workloads.
Keep your data private
Keep prompts, files, and creative assets on device for greater privacy, while local models keep workflows and context available even offline.
Customize AI for your work
Match models to specific tasks, adapt their behavior through prompts or fine-tuning, and curate trusted skills and tools around your workflow.
Optimized for Agentic AI Locally
With up to 128 GB unified memory, ProArt P16 supports LLMs with up to 120 billion parameters and a 1-million-token context window, enabling AI agents to reason, plan, and orchestrate complex creative workflows entirely on device.
RTX-Accelerated Apps, Supercharged by CUDA
RTX Spark brings native CUDA acceleration, powering creative apps, AI models, and local agentic workflows across 1,000+ accelerated apps, tools, and games.
Integrating AI Into Every Creative Workflow
Create with Generative AI
Create, edit, and render with local AI—from images and short clips to complex video projects and large 3D scenes.
Video and image creation
Unified memory: 24/32 GB
4K video editing
AI image and short-clip generation
Creative multitasking
Unified memory: 48/64 GB
Multi-camera 4K video project
AI video generation + 4K upscaling
Complex 3D and AI video1
Unified memory: 128 GB
90 GB 3D scenes and 4K AI video
4:2:2 Hardware Video Decode
Build Custom Workflows
Connect AI models and creative tools into automated workflows that generate, review, and refine images and videos locally.
Compact model workflows
Unified memory: 24/32 GB
Automatically tag and sort images
Image-to-video generation workflow
Larger, connected models
Unified memory: 48/64 GB
Train custom-style LoRAs (multi-control)
Improved visual fidelity in generations
Large model pipelines
Unified memory: 128 GB
Turn a brief into storyboard and clips
High-res video generation and upscaling
Develop Intelligent Agents
Build local AI agents that automate tasks from personal assistants to advanced multi-agent applications.
Personal AI assistant
Unified memory: 24/32 GB
7B–14B LLMs for everyday assistance
Image, file search and task automation
Agents with creative tools
Unified memory: 48/64 GB
24B–35B LLMs for coding and analysis
Agent-assisted creation (ComfyUI, Blender)
Advanced agent workflows2
Unified memory: 128 GB
70B–120B LLMs for complex research
Multiple advanced agent orchestration
ProArt AI Software Ecosystem
Local AI, built into your ProArt
MuseTree — Token-free local AI creation with FLUX and WAN
StoryCube — AI-powered scene categorization and media management
ProArt Creator Hub — Your creative portal
MuseTree
StoryCube
ProArt Creator Hub
Hybrid creation, integrated with your creative apps3
Adobe Creative Cloud — RTX-accelerated local workflows for faster AI-powered creation
Google AI Pro — Exclusive cloud integration for hybrid creativity
GoPro Premium+ — 360° video management with AI-powered scene recognition
Goodnotes — Powerful note-taking to capture ideas and spark inspiration
Google AI Pro
Adobe Creative Cloud
GoPro Premium+
Goodnotes
AI agents, working throughout your workflow
A comprehensive creative ecosystem connecting agents and apps
Hermes — A versatile AI agent that connects with ASUS MuseTree through a dedicated Hermes skill, making local generative AI creation more accessible
ComfyUI — A leading AI workflow platform offering validated, agent-ready workflows and resource packs to help lower the barriers to AI creation
Zenni Claw — An ASUS-native AI agent delivering optimized hybrid AI experiences across devices
ComfyUI
Zenni Claw
Hermes
ProArt Series
One ProArt Ecosystem for Creation and AI
ProArt brings together laptops, AI PCs, displays, components, and peripherals in a unified ecosystem built to support creators and developers across content creation, AI development, 3D rendering, and production workflows.
From personal creation to powerful AI infrastructure
Start with personal AI on ProArt RTX Spark systems, built for local creation, agentic workflows, and AI development. As workloads grow, seamlessly scale into ASUS AI infrastructure powered by NVIDIA DGX Spark and DGX Station — extending from individual creativity to shared AI development, large-model workloads, and enterprise-scale computing.
NVIDIA RTX Spark reinvents personal computing, combining personal agents, advanced content creation, and high-performance gaming in a new platform engineered from the ground up for the next wave of Windows PC experiences. Designed for creators, AI developers and gamers, RTX Spark brings NVIDIA’s full-stack AI platform and full-suite of RTX technologies to slim laptops with all-day battery life and to ultra-efficient compact desktop PCs.
Who is an RTX Spark PC designed for?
RTX Spark PCs are designed for creators, AI developers, gamers and the next generation of AI-powered workflows. Create, build, automate, and run advanced AI locally-from generative media and complex projects to agents and custom AI experiences.
What does 128GB of unified memory enable for local AI?
With 128GB of unified memory shared by the CPU and GPU, you have more room for large AI models, longer context, and workflows that combine language, vision, and generative models. This expands what you can run locally while keeping more of your creative workflow on your PC. Available capacity also depends on system and application memory usage.
Which memory configuration should I choose?
Gaming is great on all configurations of RTX Spark. Each RTX Spark has Hardware Accelerated 4:2:2 video encode and decode. The difference in memory size allows you to take on larger creative projects and use larger AI models simultaneously for agentic workflows.
• 24/32 GB: For light Video generation tasks and hybrid cloud agentic workflows this is a great start
• 48/64 GB: Recommended for 100% local genetic workloads, with additional room, larger creative tasks, and generation models
• 128 GB: Best suited for the largest local models, complex 3D scenes up to 90GB in size, and workflows combining multi-stage language, vision, and image or video
Can a ProArt RTX Spark PC run 70B- and 120B-class local LLMs?
Yes. With a suitable configuration, including 128GB unified memory, ProArt RTX Spark PCs can run selected 70B- and 120B-class LLMs locally using compatible software and appropriate quantization. Supported context length and performance vary by model architecture, precision, and other workloads running at the same time.
What is the difference between an RTX Spark PC and a GeForce RTX laptop?
RTX Spark combines the CPU and RTX GPU with a shared unified memory pool. A GeForce RTX laptop typically pairs a separate CPU with a discrete GPU that has dedicated video memory. Both accelerate AI and creative applications; RTX Spark offers greater shared memory capacity.
Can I use RTX Spark for AI agents and agentic workflows?
Yes. RTX Spark is designed to run local AI agents that can plan tasks, generate content, write code, and work across applications. With compatible agent software and configured integrations — such as APIs, MCP servers, command-line tools, and skills — you can build workflows that connect AI models to creative and productivity tools.
Why choose RTX Spark for AI agents?
RTX Spark is built for personal agents, with the performance and unified memory to run capable AI models locally, securely, and without metered cloud costs. Agents can work across the apps you already use, tap into RTX-accelerated tools on your PC, and reach the cloud when needed - all while keeping you in control of your data, access, and workflows.
What makes a ProArt RTX Spark PC different from other RTX Spark PCs?
ProArt pairs Hermes and ComfyUI to form one of the most compelling combinations for a local-first, cloud-flexible AI experience, bringing industry-validated workflows and agent-connected creator tools to users’ own PCs for generative AI creation and AI-assisted development, with the flexibility to integrate cloud services when needed. Capabilities and availability may vary by model, region, software compatibility, memory configuration, and supported integrations.
