Mobilint Launches MLD-R1 USB Edge AI Accelerator with 10TOPS Compute Power
South Korean NPU startup Mobilint has launched the MLD-R1, a USB-based edge AI accelerator powered by its REGULUS AI System-on-Chip. The device delivers 10TOPS of INT8 compute power while operating at a low thermal design power (TDP) of just 3W. This launch provides edge developers with a highly portable, low-power hardware option to run AI models locally without relying on cloud infrastructure. Its broad compatibility with major AI frameworks and host CPU architectures makes it easier to integrate into existing systems like drones and security setups. The MLD-R1 features a quad-core Arm Cortex-A53 CPU, up to 8GB of LPDDR4X memory, and a USB 5Gbps interface. It is designed for 24/7 operation with an IP65 dust and waterproof rating, and supports frameworks like PyTorch, TensorFlow, and ONNX.
## BACKGROUND
Edge AI accelerators are specialized hardware components designed to run machine learning models directly on local devices rather than in centralized data centers. Using low-precision formats like INT8 (8-bit integer) allows these chips to perform inference tasks much faster and with significantly lower energy consumption.