South Korean AI semiconductor company Mobilint has announced the MLD-R1, a compact USB-based edge AI Accelerator built for local AI inference across embedded systems, industrial devices, and development platforms. Powered by the company’s REGULUS AI SoC, the new accelerator is designed to deliver dedicated AI processing without requiring a large expansion card or high power consumption.
The MLD-R1 offers 10 TOPS of INT8 AI computing performance, making it suitable for a range of edge AI workloads including image recognition, object detection, industrial automation, smart cameras, robotics, and AI-powered IoT applications. By using a USB interface, the accelerator can be connected to supported systems quickly, allowing developers to add AI acceleration without major hardware changes.
At the core of the device is Mobilint’s REGULUS AI processor, which is paired with a quad-core Arm Cortex-A53 processor running at 1.5GHz. The accelerator is available with either 4GB or 8GB of LPDDR4X-4267 memory, giving users options based on the size of their AI models and deployment requirements.
The MLD-R1 measures 90 × 40 × 16 mm, making it compact enough for portable systems and embedded installations where space is limited. It connects through a USB interface capable of up to 5Gbps transfer speeds while maintaining a 3W thermal design power (TDP). The low power requirement makes the device suitable for continuous operation in environments where power efficiency is important.
Mobilint also highlights the durability of the MLD-R1 by giving it an IP65 rating, providing protection against dust and water. According to the company, the accelerator is designed for 24/7 operation, making it suitable for industrial equipment, factory automation, smart surveillance, and outdoor AI deployments.
On the software side, Mobilint aims to simplify AI development by supporting several widely used machine learning frameworks. Developers can build and deploy models using TensorFlow, TensorFlow Lite, PyTorch, Keras, and ONNX, reducing the need to modify existing AI workflows when moving to the new hardware.
The accelerator is compatible with both Windows and Linux, and supports multiple processor architectures, including x86, Arm, and RISC-V hosts. This broad compatibility allows the MLD-R1 to be integrated into desktops, embedded computers, industrial PCs, and custom hardware platforms.
- Mobilint introduces the MLD-R1 USB edge AI accelerator.
- Powered by the REGULUS AI SoC.
- Delivers 10 TOPS (INT8) AI computing performance.
- Features a quad-core Arm Cortex-A53 CPU at 1.5GHz.
- Available with 4GB or 8GB LPDDR4X-4267 memory.
- USB interface supports 5Gbps transfer speeds.
- Operates with a low 3W TDP.
- IP65-rated for dust and water resistance.
- Designed for 24/7 continuous operation.
- Supports TensorFlow, TensorFlow Lite, PyTorch, Keras, and ONNX.
- Compatible with Windows, Linux, x86, Arm, and RISC-V platforms.
As edge AI continues to expand across industries, compact accelerators like the MLD-R1 provide developers with another option for running AI models locally while reducing reliance on cloud-based processing. Mobilint has announced the hardware specifications of the device, although pricing and availability have not yet been disclosed.
