NVIDIA DGX Spark Personal AI Supercomputer | 128GB Unified Memory | 4TB PCIe Gen5 SSD | Grace Blackwell GB10 | Up to 1 PFLOP FP4 AI Performance | Local LLM & Generative AI Workstation

Price: ₹12,19,990 - ₹5,96,290.00
(as of Sep 08, 2026 05:40:08 UTC – Details)


NVIDIA DGX Spark is a next-generation personal AI supercomputer engineered to bring enterprise-grade AI computing to your desktop. Powered by the NVIDIA Grace Blackwell GB10 Superchip, it delivers up to **1 PFLOP of FP4 AI performance**, enabling developers, researchers, data scientists, and AI enthusiasts to build, test, and deploy advanced AI applications with exceptional speed. Featuring **128GB of high-bandwidth LPDDR5X Unified Memory**, DGX Spark can run and fine-tune AI models with up to **200 billion parameters locally**, eliminating the need for costly cloud infrastructure while keeping sensitive data on-premises. The integrated **4TB PCIe Gen5 NVMe SSD** provides ultra-fast storage for large AI models, datasets, and development environments. Equipped with **6,144 CUDA Cores**, **5th Generation Tensor Cores**, **4th Generation RT Cores**, Wi-Fi 7, 10Gb Ethernet, and NVIDIA’s complete AI software stack including CUDA, TensorRT, and NVIDIA NIM, DGX Spark is purpose-built for generative AI, large language models (LLMs), machine learning, computer vision, robotics, scientific computing, and edge AI. Its compact, energy-efficient design delivers data center-class AI performance in a desktop form factor, making it the ideal solution for professionals who demand powerful local AI computing without compromise.
PERSONAL AI SUPERCOMPUTER – Experience data center-class AI performance in a compact desktop form factor with the NVIDIA DGX Spark, purpose-built for AI developers, researchers, data scientists, and enterprises.
UP TO 1 PFLOP AI PERFORMANCE – Powered by the NVIDIA Grace Blackwell GB10 Superchip with 5th Generation Tensor Cores, delivering up to 1 PetaFLOP (FP4) of AI compute for generative AI, machine learning, and deep learning workloads.
RUN 200B PARAMETER AI MODELS LOCALLY – The massive 128GB LPDDR5X Unified Memory enables local inference and fine-tuning of AI models with up to 200 billion parameters, reducing cloud dependency while keeping your data private.
128GB HIGH-BANDWIDTH UNIFIED MEMORY – Coherent CPU-GPU memory architecture allows seamless data sharing without copying between system RAM and VRAM, accelerating AI workflows and improving efficiency.
ULTRA-FAST 4TB PCIe GEN5 NVMe SSD – High-performance Gen5 storage delivers lightning-fast boot times, rapid model loading, and ample space for AI models, datasets, virtual machines, and development projects.
BUILT FOR GENERATIVE AI & LLMs – Optimized for Large Language Models (LLMs), AI agents, Retrieval-Augmented Generation (RAG), computer vision, robotics, coding assistants, image generation, and multimodal AI applications.
NVIDIA AI SOFTWARE STACK READY – Supports CUDA, TensorRT, TensorRT-LLM, NVIDIA NIM, PyTorch, RAPIDS, Triton Inference Server, and other industry-leading AI frameworks for accelerated development and deployment.
6,144 CUDA CORES & ADVANCED GPU ARCHITECTURE – Leverage NVIDIA Blackwell architecture with 6,144 CUDA Cores, 5th Gen Tensor Cores, and 4th Gen RT Cores for exceptional AI acceleration, parallel computing, and graphics performance.
HIGH-SPEED ENTERPRISE CONNECTIVITY – Equipped with Wi-Fi 7, Bluetooth 5.4, 10 Gigabit Ethernet, NVIDIA ConnectX-7 networking, HDMI 2.1a, and USB Type-C ports for fast networking and seamless integration into professional AI environments.
COMPACT, ENERGY-EFFICIENT & FUTURE-READY – Designed to deliver workstation-class AI computing in an ultra-compact desktop system, offering exceptional performance per watt for AI development, scientific computing, digital twins, edge AI, and enterprise innovation.

Product summary

shift
+
opt
+
D