CN JP

LM2-100

25 TOPS AI Performance
AVG.5W
2 GB LPDDR4x
25 TOPS AI Performance
25 TOPS AI Performance
The LM2-100 unlocks up to 25 TOPS of AI computing power, enabling real-time inference for advanced neural networks such as YOLO, ResNet, and Vision Transformers. The AI accelerator module can effortlessly handle multi-stream video analytics and complex edge AI workloads with high efficiency.
Ultra-Low Power at Just 5W
Ultra-Low Power at Just 5W
LM2-100 is specifically designed for energy efficiency, with a typical power consumption of just 5W. It seamlessly integrates with various fanless industrial computers, AI edge controllers, and other devices. At full load, its energy efficiency ratio reaches up to 5 TOPS/W, allowing customers to achieve full performance AI inference on compact devices without the need to redesign high-power power supply systems.
Standard M.2 2242 Form Factor
Standard M.2 2242 Form Factor
The LM2-100-VO adopts the standard M.2 2242 B+M-KEY form factor and connects to the host through a PCIE GEN3X2 interface, facilitating integration into embedded devices such as industrial PCs and single-board computers without requiring additional structural space.
Robust & Developer-Friendly <br />Software Ecosystem
Robust & Developer-Friendly
Software Ecosystem
Accelerate deployment with a comprehensive SDK that includes model conversion tools, optimized libraries, and ready-to-use reference designs. With native support for TensorFlow, PyTorch, and ONNX, plus access to a rich pre-trained model zoo, bringing AI applications to life has never been easier.
I/O port introduction
NPU DeepX M1M
Memory 2GB LPDDR4x
AI Performance 25 TOPS
I/O Interface M.2 Key B+M
Host Interface PCle 3.0X2
Power Requirement 3.3V upto 3A
PCB 22 x 42 x 4.1mm (0.86” x 1.65” x 0.16”)
Operating System Windows, Linux
Operating Temperature -25°C ~ 65°C 
Storage Temperature -25°C ~ 70°C 
AI Frameworks Supports TensorFlow, TensorFlow Lite, ONNX, Keras, and PyTorch, converted via the Dataflow compiler.

LM2-100-User-Manual
1.77 M
2026-03-23
Download

Inquiry

Inquiry Download Technical Support RMA Events