Skip to main content

Radxa NX5

Product Overview

The Radxa NX5 is a high-performance compute module based on the Rockchip RK3588S, integrating a 4-core Cortex-A76 + 4-core Cortex-A55 CPU, ARM Mali-G610 MP4 GPU, and 6 TOPS@INT8 NPU, supporting mainstream deep learning frameworks. It offers rich I/O and expansion capabilities with 8K video codec support, suitable for edge computing, machine vision, smart terminals, multimedia processing, and more.

The Radxa NX5 can be paired with the official IO board for quick feature validation and prototyping, helping users complete power-on debugging, interface evaluation, and application development in a short time.

Product Images

Specifications

ModelRadxa NX5
SoCRockchip RK3588S
CPU4-core Cortex-A76 + 4-core Cortex-A55
GPUARM Mali-G610 MP4
- Supports OpenGL ES 1.1, 2.0 and 3.2, OpenCL 1.1, 1.3 and 2.2, Vulkan 1.1 and 1.2
- High-performance 2D image acceleration module embedded
NPU6 TOPS@INT8
- Supports INT4 / INT8 / INT16 / FP16 / BF16 / TF32 compute precision
- Supports TensorFlow, Caffe, Tflite, Pytorch, Onnx NN, Android NN and other deep learning frameworks
MemoryLPDDR4X
- Capacity: up to 16GB
Video codecVideo encoding
- Up to 8K@30fps H.265 / H.264 encoding
Video decoding
- Up to 8K@60fps AV1 / AVS2 / VP9 / H.265 / H.264 decoding
StorageOptional onboard eMMC
- eMMC 5.1, up to 512GB
Onboard SPI Flash (optional)
1x SDMMC
Networking1x Gigabit Ethernet
Display1x HDMI TX
- Up to 8K@60Hz
1x eDP TX
- Up to 4K@60Hz
1x DP TX
- Up to 8K@30Hz
1x MIPI DSI (2-lane)
Camera2x MIPI CSI (2-lane) or 1x MIPI CSI (4-lane)
1x MIPI CSI (4-lane)
1x MIPI CSI (2-lane)
USBUSB 3.1 Gen1
USB 2.0
PCIePCIe2.0
Other interfacesSupports UART, I2C, I2S, CAN, PWM, GPIOs and more
Connector260-pin SO-DIMM connector
OSSupports Debian, Yocto, Buildroot, Android 14
Dimensions70 mm x 45 mm

Interface Description

No.DescriptionNo.DescriptionNo.Description
1LPDDR4X2Rockchip RK3588S3Onboard eMMC
4260-pin SO-DIMM5SPI Flash

Application Scenarios

Edge Computing and Smart Gateways

With a 6 TOPS NPU and Gigabit Ethernet, it can perform AI inference tasks such as video analytics, object detection, and face recognition locally without relying on the cloud. This enables edge-side intelligent decision-making and data pre-processing, significantly reducing network bandwidth and cloud computing pressure.

Machine Vision and Industrial Automation

It can be used for machine vision scenarios such as visual inspection, recognition, and measurement. With support for CAN bus and other industrial interfaces, it can work with fieldbuses and industrial control equipment to build flexible production lines, automated inspection stations, and smart devices.

Multimedia Processing and Smart Displays

Supports 8K@60fps video decoding and 8K@30fps video encoding, suitable for digital signage, video surveillance, video conferencing, smart advertising displays, information kiosks, self-service devices, smart retail terminals, and other multimedia applications. High-resolution display output combined with local intelligent analysis enables richer human-machine interaction experiences.

Service and Mobile Robots

The 8-core CPU combined with NPU offers various sensor expansion interfaces and wired networking capabilities. It can serve as a main controller platform for mobile or collaborative robots, used for path planning, environmental perception, target tracking, and remote operations and status monitoring.

Education, Training, and R&D Validation

Suitable as an embedded and AI teaching practice platform for universities and training institutions. It is also ideal as a basic hardware platform for enterprise solution prototyping and functional validation, accelerating the process from concept design to product launch.

    You need to be logged into GitHub to post a comment. If you are already logged in, please ignore this message.

    Radxa-docs © 2026 by Radxa Computer (Shenzhen) Co.,Ltd. is licensed under CC BY 4.0