




product
Product Specifications
Mini AI supercomputer designed for local deployment ensuring data security
Supports smooth local operation of 120B parameter AI models
AI office assistant with intelligent collaboration, knowledge Q&A, and skill expansion
Combines NVIDIA GB10 Grace Blackwell superchip with NVIDIA AI software stack
Provides comprehensive AI development and deployment solutions
Compact design with advanced AI tools and NVIDIA ConnectX-7 networking technology
Supports prototyping, fine-tuning, and inference of large AI models up to 200 billion parameters
High-performance 20-core ARM v9.2-A CPU (NVIDIA GB10 Grace)
NVIDIA Blackwell GPU integrated with 128 GB unified memory
NVLink™ M-C2C technology for consistent CPU+GPU memory model
Fifth-generation Tensor Cores supporting FP4 precision
AI performance up to 1 petaFLOP (1000 TOPS)
128 GB LPDDR5x unified system memory
High bandwidth up to five times PCIe Gen5

AMD Ryzen AI Max+ 395
CN¥14,989.00

i914900K (24 cores, 3.25.6 GHz)
CN¥22,799.00
Supports fine-tuning of AI models with up to 70 billion parameters
M.2 2230/2242 NVMe PCIe 5.0 x4 (backward compatible with PCIe 4.0 x4)
AW-EM637 Wi-Fi 7 (Gig+) 2x2 + Bluetooth 5
1 × 10G LAN port
1 × ConnectX CX-7 100 Gbps (2x QSFP) for high-speed networking and system chaining
IEEE 1588 V2 PTP support for microsecond-level time synchronization
1 × Power button
3 × USB 3.2 Gen 2x2 Type-C (20 Gbps) with DisplayPort 2.1 alternate mode
1 × USB 3.2 Gen 2x2 Type-C with 180W EPR PD 3.1 power delivery input
1 × HDMI 2.1b port
1 × ConnectX CX-7 100 Gbps (2x QSFP)
1 × 10G LAN port
1 × Kensington security lock slot
Optimized thermal design for stable AI performance under high load
Compact form factor delivering high-density AI computing power
NVIDIA DGX SPARK AI software stack including NVIDIA NGC and Blueprints
Supports PyTorch, Jupyter, Ollama for prototyping and inference
NVIDIA RAPIDS for accelerated end-to-end data science workflows
Supports AI frameworks for edge applications including ISAAC, METROPOLIS, HOLOSCAN
Enables local AI model development, testing, fine-tuning, and deployment
Easy migration to NVIDIA DGX Cloud or other NVIDIA accelerated data centers
Local AI development and flexible deployment
Prototyping, testing, and validating AI models and applications
AI research, data science, and edge computing applications
Supports large parameter generative AI models such as LLAMA 3.1
Kensington lock slot for physical security
PD input supporting 180W EPR PD 3.1 specification