The AIBOX-9075 is an industrial Edge AI box built around the Qualcomm IQ-9075 SoC, designed for running AI workloads directly on the device without relying on the cloud. With up to 200 TOPS of AI performance, it is designed for edge computing, private LLM deployment, robotics, computer vision, and more.
The system comes with 36GB LPDDR5 ECC memory, 128GB UFS 2.2 storage, and an M.2 PCIe 4.0 NVMe SSD slot. Connectivity options include dual 2.5GbE with TSN, optional Wi-Fi 6, Bluetooth 5.2, and 4G/5G, while peripheral interfaces include eight GMSL2 camera inputs, HDMI 2.0, USB 3.0, CAN-FD, RS485, and opto-isolated digital I/O. Its rugged aluminum enclosure operates from -40°C to 85°C, making it suitable for continuous deployment in industrial and outdoor environments.
Firefly AIBOX-9075 specifications:
- SoC – Qualcomm DragonWing IQ-9075
- CPU
- Octa-core Kryo Gen 6 (Cortex-A78C-based) application cores @ up to 2.36 GHz
- Quad-core Cortex-R52 real-time cores @ up to 1.85GHz
- GPU – Adreno 663 GPU delivering up to 1.2 TFLOPS FP32 with secure GPU compute; supports Vulkan 1.2, OpenGL ES 3.2, OpenCL 2.0 FP, Adreno NN Direct
- VPU – Adreno VPU 765
- Video Decode:
- AV1 / HEVC / H.265 / H.264 / VP9 / MPEG-2
- 1x 8Kp60 / 2x 8Kp30 / 4x 4Kp60 / 8x 4Kp30 / 16x 1080p60 / 32x 1080p30
- Video Encode:
- H.264 / H.265 / HEIF / HEIC
- 2x 4Kp60 / 4x 4Kp30 / 8x 1080p60 / 16x 1080p30
- Concurrent 2x 4Kp60 encode + 2x 4Kp60 decode
- Video Decode:
- AI accelerator
- Hexagon Tensor Processor (HTP) with quad HVX and dual HMX
- 200 TOPS @ INT8 (Sparse Computing Power)
- Supports deep-learning frameworks: TensorFlow, PyTorch, ONNX, Paddle, Caffe, DarkNet, etc.
- ISP – Qualcomm Spectra 690 (2x IFE + 5x IFE_L)
- DPU – Dual Adreno DPU1199
- CPU
- System Memory – 36GB LPDDR5 with ECC support
- Storage
- 128GB UFS 2.2 flash
- M.2 2280 (PCIe 4.0 x4) socket for NVMe SSD
- Video I/O
- Output – HDMI 2.0 up to 4K @ 60Hz
- Input – 8x GMSL2 via 2x 4-pin Mini FAKRA connectors
- Networking
- 2x 2.5GbE RJ45 ports
- Optional Wi-Fi 6 and Bluetooth 5.2 via M.2 E-key slot
- Optional 4G/5G cellular connectivity via mini PCIe slot plus SIM card slot
- Mounting holes for 4x 5G/4G cellular and 2x Wi-Fi antennas
- USB – 2x USB 3.0 Type-A (Max 1A), 1x USB Type-C for debugging
- Expansion – 36-pin 3.5mm pitch Phoenix terminal block with 2x RS485, 2x CAN-FD, 12x Opto-isolated DI/DO
- Misc – Power and Burn buttons; Working, Flash, and Power LEDs
- Power
- Supply – 12V/5A via 5.5 x 2.1mm DC jack
- Consumption – 1.08W (sleep), 12W (normal), 60W (max)
- Dimensions – 110.2 x 102.25 x 72.0 mm (excluding mounting ears)
- Temperature – Operating: -40°C to 85°C; Storage: -40°C to 90°C
- Humidity: 10% to 90% RH (non-condensing)
Firefly says the AIBOX-9075 runs Ubuntu and Yocto Linux using the Qualcomm Linux software stack. It supports the QNN AI inference framework, as well as TensorFlow, TensorFlow Lite, PyTorch, ONNX, Keras, and Caffe. It also supports DeepSeek-R1, Gemma, Llama, Qwen, and Phi, as well as Stable Diffusion for image generation. The company also mentions that the Hexagon NPU delivers up to 22 tokens/s with Llama2-7B and up to 12 TPS with 13B-parameter models. More information, including the getting-started guide, power-on instructions, debug console setup, AI development tutorials (AidLux, AidLite, AidGen, and AidStream), serial, watchdog, and RTC usage, as well as software resources, is available on the wiki.
The IQ-9075-powered Firefly AIBOX-9075 Edge AI box can be purchased on AliExpress for $1,600.69 shipped (reseller), or directly on the Firefly store for $1,239.00. The company also offers optional accessories, including the WXT14R2201 Wi-Fi module ($15.00), EC20 4G-LTE module ($39.00), and a FORESEE M.2 2280 SSD ($79.00).
Debashis Das is a technical content writer and embedded engineer with over five years of experience in the industry. With expertise in Embedded C, PCB Design, and SEO optimization, he effectively blends difficult technical topics with clear communication
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress. We also use affiliate links in articles to earn commissions if you make a purchase after clicking on those links.
