CN

A gumstick-sized M.2 2280 card delivering M50's full performance for local 10B+ parameter LLM inference with fanless operation, multi-OS support and full-stack development tools - ideal for AI PCs/Sticks in consumer, office and industrial applications requiring private, low-latency AI.

Specifications
Dimensions
22×80mm
Interfaces
PCIe Gen4 ×4, M.2 M-Key
AI Accelerator
1× MOMAGIC ® 50
Physical Computing Power
160 TOPS @INT8
100 TFLOPS @bFP16
Typical power consumption
≤ 13W
Memory Capacity
12 GB 192-bit LPDDR5, supporting a maximum of 48GB
Memory Bandwidth
153.6 GB/s
System Support
Win11 / Linux / Android
Software Support
Houmo Dadao™️ Software Platform
Power Management
Multi-level power consumption modes, ultra-low power standby as low as 4mW
Operating Temperature
0-50℃
Cooling Solution
Custom heatsink, supporting active cooling / passive cooling
Specifications
Dimensions
22×80mm
Interfaces PCIe
Gen4 ×4, M.2 M-Key
AI Accelerator
1× MOMAGIC ® 50
Physical Computing Power
160 TOPS @INT8, 100 TFLOPS @bFP16
Typical power consumption
≤ 13W
Memory Capacity
12 GB 192-bit LPDDR5, supporting a maximum of 48GB
Memory Bandwidth
153.6 GB/s
System Support
Win11 / Linux / Android
Software Support
Houmo Dadao™️ Software Platform
Power Management
Multi-level power consumption modes, ultra-low power standby as low as 4mW
Operating Temperature
0-50℃
Cooling Solution
Custom heatsink, supporting active cooling / passive cooling