Skip to main content

Cubie Series

The Cubie series is a line of single-board computers from Radxa based on Allwinner SoCs.

Model Comparison

ModelSoC / CPUNPU PerformanceMemory OptionsSystem / Onboard StorageStorage Expansion & BusesNetworking & WirelessDisplay & CameraTypical I/O
Cubie A5EAllwinner A527 / T527, 8× Cortex-A55, up to 2.0GHzT527 variant: 2 TOPSLPDDR4/4x: 1GB / 2GB / 4GBSPI Flash, onboard eMMC (optional 0–32GB)microSD, M.2 M-Key NVMe (2230)Dual Gigabit Ethernet (one PoE via PoE HAT), Wi‑Fi 6, Bluetooth 5.4HDMI 4K@60, 1× MIPI DSI, 1× MIPI CSI40-pin GPIO, USB 3.0 Type‑A, USB‑C OTG
Cubie A7AAllwinner A733, 2× A76 + 6× A55, up to 2.0GHz3 TOPSLPDDR5: 2GB / 4GB / 6GB / 8GB / 12GB / 16GB8MB SPI NOR FlashmicroSD, eMMC module, UFS moduleGigabit Ethernet (PoE via PoE HAT), Wi‑Fi 6, Bluetooth 5.4HDMI 4K@60, 1× MIPI DSI, 1×/2× MIPI CSI40-pin GPIO, PCIe 3.0 FPC, rich USB
Cubie A7ZAllwinner A733, 2× A76 + 6× A55, ultra-compact 65×30mm board3 TOPS@INT8LPDDR4/4x: 1GB / 2GB / 4GB / 8GB / 16GBOptional onboard UFS (up to 1TB)microSD, PCIe 3.0 FPC (1-lane)Onboard Wi‑Fi 6, Bluetooth 5.4Micro HDMI 4K@60, USB‑C 3.0 DP Alt, MIPI CSI40-pin GPIO, dual USB‑C

Model Highlights and Selection Advice

Cubie A5E: Dual-network and industrial-first

  • Dual Gigabit Ethernet with one PoE-capable port (requires PoE HAT), suitable for industrial gateways, edge routers, and dual-network isolation.
  • Optional T527 SoC with 2 TOPS NPU, adequate for lightweight edge AI inference (simple detection/classification).
  • M.2 M-Key NVMe support enables compact high-speed local storage expansion.
  • Standard 40-pin GPIO, HDMI + MIPI DSI / CSI cover typical industrial control and HMI needs.
  • Best for: projects prioritizing dual-network/PoE and industrial gateway use cases where board size is not extremely constrained.

Cubie A7A: General-purpose high-performance edge AI platform

  • Allwinner A733 with 3 TOPS NPU makes it suitable for general edge AI and desktop/multimedia development.
  • LPDDR5 up to 16GB supports larger model inference, multitasking, and desktop graphics workloads.
  • Supports eMMC / UFS modules so system storage can be flexibly chosen for production.
  • PCIe 3.0 FPC enables connecting NVMe, AI accelerator cards, or other high-speed peripherals for strong expandability.
  • Rich USB, audio interfaces and 40-pin GPIO balance desktop experience and embedded extensibility.
  • Best for: projects needing a balance of performance and expansion, with some space for the board.

Cubie A7Z: Ultra-compact edge AI module

  • Ultra-small 65×30mm board offering A7A-like CPU/NPU performance in a tiny form factor.
  • Optional onboard UFS (up to 1TB) provides high-capacity, high-performance system storage without extra modules.
  • Dual USB‑C (one with USB 3.0 + DP Alt), suitable for compact enclosures or modular products.
  • Onboard Wi‑Fi 6 / Bluetooth 5.4, 40-pin GPIO and PCIe 3.0 FPC keep expansion capabilities despite small size.
  • Best for: space-constrained designs requiring onboard high-speed storage and high integration for edge AI or embedded products.

    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