When you try to add AI power to a robot or a smart camera, you often face two problems. The first is that most boards are big, need a lot of power, and are hard to set up. The second is that they lack the right software tools, so you waste time just getting them to work. This matters because a slow or unstable system can stop a project in its tracks. The Firefly NVIDlA Jetson Orin Mainboard gives you a small, low-power board that already includes Ubuntu, CUDA and the latest AI libraries, so you can start testing your ideas right away.
Firefly NVIDlA Jetson Orin Mainboard, AIO-OrinNano (8GB), Hexa-core 64-bit CPU with Firefly
Ready to turn your ideas into real AI devices? The Firefly NVIDlA Jetson Orin Mainboard, AIO-OrinNano (8GB), Hexa-core 64-bit CPU with Firefly is a complete edge-compute kit that fits in the palm of your hand. It is built for developers who need fast inference, vision processing and robot control without building a whole PC. If you have struggled with noisy performance or missing drivers, this board solves those issues by delivering 67 TOPS of AI power, a 1024-core Ampere GPU and a ready-to-run Ubuntu 22.04 system.
Key Features
- Hexa-core ARM Cortex-A78AE CPU up to 1.7 GHz
- 1024-core NVIDIA Ampere GPU with 32 Tensor Cores
- 8 GB LPDDR5 memory for smooth multitasking
- 67 TOPS AI performance for vision and generative models
- One M.2 M-Key slot for PCIe NVMe SSD storage
- HDMI 2.0 output up to 4K @ 30 Hz
- Two MIPI CSI-D-PHY camera interfaces (4-lane or 2-lane)
- 12 V to 24 V wide input range, 30 W max power
- Rich expansion: GPIO, IC, SPI, UART, CAN, RS485, SIM slot
- Three buttons (Reset, Recovery, Power) for easy maintenance
Specifications
| AI Performance |
Supports ROS robot model. Supports Ollama for local LLM deployment. Compatible with Llama 3, Phi-3 Mini, EfficientViT,
NanoOWL, NanoSAM, SAM, TAM, Flux, Stable Diffusion, and Stable Diffusion XL for AI, vision, and image generation applications.
|
| AI Software Stack |
NVIDIA Jetson Orin software stack with Ubuntu 22.04, CUDA, TensorRT, cuDNN, and MLPerf support for Transformer-based
generative AI inference.
|
| CPU |
Hexa-core 64-bit ARM Cortex-A78AE v8.2, up to 1.7 GHz |
| AI Performance (TOPS) |
67 TOPS |
| GPU |
1024-core NVIDIA Ampere GPU with 32 Tensor Cores |
| Memory |
8GB LPDDR5 |
| Storage |
1 × M.2 M-Key slot for PCIe NVMe SSD (2242/2260/2280) |
| Power Input |
DC 12V (5.5 × 2.1 mm), supports 9V–24V wide voltage input |
| Power Consumption |
Normal: 6W (12V/500mA); Maximum: 30W (12V/2500mA) |
| Video Output |
1 × HDMI 2.0 (up to 4K @ 30Hz) |
| Video Input |
2 × MIPI CSI D-PHY (1 × 4-lane or 2 × 2-lane), Line-in via pin header |
| Audio |
1 × 3.5 mm audio jack with microphone input (CTIA standard) |
| Operating System |
Ubuntu 22.04 (Jetson Linux) with NVIDIA CUDA, TensorRT, cuDNN, and graphics acceleration |
| Dimensions |
122.89 × 85.04 × 35.28 mm |
| Weight |
129 g (without fan), 180 g (with fan) |
| Operating Temperature |
-20°C to 60°C |
| Storage Temperature |
-20°C to 70°C |
| Storage Humidity |
10%–90% RH (non-condensing) |
| Expansion Interfaces |
1 × 4-pin fan header, 1 × SIM card slot, 1 × 3-pin debug header, 20-pin GPIO header (USB 2.0, SPI,
2 × I²C, Line-in, Line-out, GPIO), 8-pin Phoenix connector (RS485, RS232, CAN 2.0)
|
| Buttons |
Reset, Recovery, Power |
