Specifications
AI Performance
157 TOPS (INT8)
SoC
NVIDIA Jetson Orin NX 16GB module
CPU
Octa-core 64-bit Arm Cortex-A78AE v8.2, up to 2.0 GHz
GPU
1024-core NVIDIA Ampere GPU, 32 Tensor Cores
Memory
16GB LPDDR5
Storage
128GB PCIe NVMe SSD (M.2, configurable)
Video Decode
1x 8K30 / 2x 4K60 / 4x 4K30 / 9x 1080p60 / 18x 1080p30 H.265
Video Encode
1x 4K60 H.265
USB
2x USB 3.0 Type-A (max 1A)
Display
1x HDMI 2.1, up to 4K@60fps
Power Input
DC 12V (5.5 x 2.1mm)
Power Consumption
7.2W typical / 33.6W max
Watchdog
External hardware watchdog supported
Enclosure
Industrial full-metal aluminium shell, grille top cover
Dimensions
93.4 x 93.4 x 50 mm
Weight
Approx. 500 g
Software
Ubuntu 22.04 Jetson BSP, CUDA, TensorRT, cuDNN, JetPack
Certifications
CE / FCC · 1-year Firefly warranty
Overview
The Firefly AIBOX-OrinNX is a cube-shaped, fanless edge AI computer powered by the NVIDIA Jetson Orin NX 16GB module, delivering up to 157 TOPS of INT8 AI performance. Its full-metal aluminium enclosure measures just 93.4 x 93.4 x 50 mm and weighs roughly 500 g, with a grille top cover engineered for sustained heat dissipation under high-temperature, always-on workloads.
At its core the AIBOX-OrinNX combines an eight-core Arm Cortex-A78AE CPU running up to 2.0 GHz with a 1024-core NVIDIA Ampere GPU and 32 Tensor Cores, backed by 16GB of LPDDR5. A 128GB PCIe NVMe SSD ships as standard, and hardware H.265 decode supports up to 1x 8K30 or 18x 1080p30 streams - ideal for multi-camera inspection and video analytics. A single HDMI 2.1 port outputs 4K@60fps for local visualisation.
The system runs an Ubuntu 22.04-based Jetson BSP with CUDA, TensorRT and cuDNN, enabling private on-device deployment of large language models (LLaMA2, ChatGLM, Qwen), vision models (ViT, Grounding DINO, SAM) and AIGC image generation. An external hardware watchdog and Docker containerisation make it well suited to unattended industrial and robotics deployments.
Key Benefits
157 TOPS of Jetson Orin NX compute in a 93.4mm fanless cube. 128GB NVMe storage as standard. 8K30 hardware decode for multi-stream vision. External watchdog for reliable unattended operation. Full JetPack stack with Docker for fast model deployment.
Applications
Multi-camera vision analytics, robotics and ROS2 development, industrial inspection, smart retail and access control, edge LLM/VLM inference, and portable inspection devices.
Request a Quote — FIREFLY AIBOX-ORINNX — 157 TOPS JETSON ORIN NX FANLESS EDGE AI COMPUTER
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →