🚀 Radxa ROCK 4D RK3576 8-Core Edge AI Board

6 TOPS NPU, RKNN & LLM Support for Industrial and Edge AI

Important: The current version of ROCK 4D cannot boot from eMMC. Please boot from microSD or supported storage instead.Key Features💪 8-Core RK3576 Processor

Powerful octa-core CPU designed for industrial AI and edge computing workloads.

🧠 6 TOPS NPU

Dedicated Neural Processing Unit with RKNN support for on-device AI inference.

🤖 LLM Support

Capable of running local large language model workloads at the edge.

⚙️ Multiple RAM Options

Available in 2GB, 4GB, 6GB, 8GB, 12GB and 16GB configurations to match your project needs.

Choosing Your Option

This listing offers the Radxa ROCK 4D single-board computer in six RAM capacities: 2GB, 4GB, 6GB, 8GB, 12GB and 16GB. All variants share the same RK3576 8-core CPU, 6 TOPS NPU, and board layout. Higher RAM options are recommended for demanding AI/LLM workloads, while lower RAM options suit lighter edge computing tasks.

SpecificationsModelRadxa ROCK 4DProcessorRK3576 8-Core CPUAI Performance6 TOPS NPU with RKNN supportAI/LLM CapabilitySupports local LLM inferenceRAM Options2GB / 4GB / 6GB / 8GB / 12GB / 16GBBoot NoticeCannot currently boot via eMMCTypical ApplicationEdge Computing and AI ApplicationsOriginMainland ChinaProduct ImagesFAQCan this board boot from eMMC?

No, the current version of the ROCK 4D cannot currently boot via eMMC. Please use a microSD card or another supported boot method.

What RAM options are available?

This board is available in 2GB, 4GB, 6GB, 8GB, 12GB and 16GB configurations.

Is this suitable for AI projects?

Yes, it includes a 6 TOPS NPU with RKNN support and can run local LLM workloads, making it well suited for edge AI and industrial applications.