NVIDIA Jetson T5000-based Firefly EC-ThorT5000 industrial PC features 8x GMSL2 camera interfaces, 4x 10GbE RJ45 ports

The Firefly EC-ThorT5000 is an industrial edge AI PC built around NVIDIA’s new Jetson T5000 Thor series module, with up to 2070 TFLOPS FP4 performance. The system is one of the first commercial products based on the Jetson Thor platform, launched in late August 2025.

The EC-ThorT5000 provides eight GMSL2 camera inputs, four 10GbE and four GbE (PSE) ports, dual NVMe storage, HDMI 2.0, USB 3.0, CAN/RS485/RS232, SIM support, and expansion options for 5G, 4G, and WiFi 6. It also offers up to 92-channel 1080p video decoding in a rugged all-aluminum chassis designed for 24/7 operation, making it suitable for robotics, smart city deployments, and high-density edge AI workloads.

Firefly EC-ThorT5000 specifications:

• SoM – NVIDIA Jetson Thor T5000 module
  • CPU – 14-core 64-bit ARM Neoverse-V3AE @ up to 2.6GHz
  • GPU – 2560-core Blackwell GPU with 96 Tensor Cores, 10-TPC MIG
  • VPU
    • Video Decode – 4x 8Kp30 (H.265), 10x 4Kp60 (H.265), 22x 4Kp30 (H.265), 46x 1080p60 (H.265), 92x 1080p30 (H.265), 82x 1080p30 (H.264), 4x 4Kp60 (H.264)
    • Video Encode – 6x 4Kp60 (H.265), 12x 4Kp30 (H.265), 24x 1080p60 (H.265), 50x 1080p30 (H.265), 48x 1080p30 (H.264), 6x 4Kp60 (H.264)
  • AI performance – 2070 TFLOPS FP4 (Sparse)
  • Memory – 128 GB 256-bit LPDDR5X 273 GB/s
• Storage – 2x M.2 M-Key socket for NVME SSDs
• Video Output – 2x HDMI 2.0 up to 4Kp60
• Camera– 8x GMSL2 (Gigabit Multimedia Serial Link 2) via 2x FAKRA connectors
• Audio – 3.5mm audio jack (Support MIC recording, CTIA)
• Networking
  • Ethernet
    • 4x 10GbE RJ45 ports
    • 4x GbE RJ45 ports with PSE/PoE support
  • Wireless (via expansion)
    • WiFi 6 / BT 5.2 via M.2 E-Key socket
    • Optional 4G LTE via Mini PCIe socket and SIM card slots
    • Optional 5G via M.2 B-Key socket
  • GNSS – Built-in GPS module
  • Antennas – 3x 5G, 1x 4G/5G, 1x GPS, 2x WiFi/Bluetooth
• USB
  • 4x USB 3.0 Type-A ports
  • 2x Type-C USB ports with debug and OTG support
• Expansion
  • 2x M.2 M-Key 2280 PCIe sockets for NVMe SSD, or other M.2 PCIe modules
  • M.2 B-Key socket for 5G cellular connectivity
  • Mini PCIe socket for 4G LTE support
  • M.2 E-Key socket for WiFi 6 / BT 5.2
• Misc
  • Power, Reset, and Recovery buttons
  • Run and Power LED indicator
• Power Supply – 24V DC (9–36V supported) via 5.5×2.1mm power barrel jack
• Dimensions – 277.95 × 136.09 × 88.0 mm
• Temperature Range – Operating: -20°C to 60°C; storage: -20°C to 70°C
• Humidity – 10 ~ 90%, non-condensing

The Firefly EC-ThorT5000 supports the Jetpack SDK based on Ubuntu 24.04 with full support for CUDA, TensorRT, and cuDNN. It supports deep-learning frameworks including PyTorch, TensorFlow, PaddlePaddle, MXNet, MATLAB, and Keras, along with Docker for easy containerized deployment. The system also works with Ollama and ComfyUI for local LLMs and image-generation workflows, and can privately deploy popular models such as Deepseek-R1, Llama, Qwen, EfficientVIT, SAM, and Stable Diffusion. At this point, I would love to say that more information can be found on the company’s wiki, but at the time of writing, the link to the wiki is not up.

Appearance-wise, the EC-ThorT5000 looks very similar to the Firefly EC-AGXOrin, and it appears that Firefly reused their proven EC-series rugged enclosure for rapid time-to-market, with only the interface cutouts differing, but performance-wise, the EC-ThorT5000 is in a completely different league. NVIDIA only released the Jetson AGX Thor Developer Kit back in August this year. More Jetson T5000-based commercial products are coming soon, and the Firefly EC-ThorT5000 is just the first one we’ve covered.

Pricing is not yet available, but given the specifications, this type of industrial-grade Jetson T5000 embedded PC could very well end up as a $3,000+ product. A few more details may be available on the product page.

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.