The platform combines an ARM-based Grace CPU, Blackwell RTX GPU and up to 128GB of unified memory for local AI, development, content creation and gaming.
Two N1X Configurations
| N1X High-End | N1X Laptop | |
| CPU | 20-core Grace | 18-core Grace |
| GPU | 6,144-core Blackwell RTX | 5,120-core Blackwell RTX |
| Memory | 24GB–128GB unified | 24GB–32GB unified |
| Devices | Laptops + desktops | Laptops |
Up to 128GB shared between the CPU and GPU removes the VRAM ceiling of conventional discrete GPUs and allows significantly larger AI models to run locally.
Built for Local AI
NVIDIA positions RTX Spark for:
- Local LLMs with up to 120B parameters
- AI agents running directly on the PC
- CUDA-based AI development
- Local inference without cloud APIs
- 12K 4:2:2 video editing
- Ray tracing and DLSS
- 1440p gaming at 100+ FPS
A 120B model will still require quantization to fit into 128GB, but the memory capacity enables workloads that would otherwise require high-memory GPUs, multi-GPU systems or cloud instances.
Developers can use RTX Spark to prototype LLM applications, RAG systems, coding agents and multimodal tools locally, reducing API costs and latency.
Because Grace is ARM-based, Windows ARM compatibility will be an important consideration for native libraries, Python packages and other x86-dependent development tools.
First RTX Spark Hardware
Lenovo unveiled the Yoga 9n 2-in-1, while Acer demonstrated a compact RTX Spark desktop at IFA 2026. The first devices are scheduled to ship in October 2026.
