Yuan EDG6N0-S T5X NVIDIA Jetson Thor T5000/T4000 Edge AI Box PC with QSFP28, 5GbE and Dual 4K Displays

Thor-based AI often moves into production once a prototype has proved its models, and at that point an enclosed unit with fast networking usually makes deployment simpler. The EDG6N0-S T5X from Yuan is an edge AI box PC built on the NVIDIA Jetson Thor T5000 or T4000 module, with QSFP28 for high-bandwidth sensor or storage links and two 4K60 display outputs.

The T5000 gives up to 2070 FP4 TFLOPS from a 14-core Neoverse-V3AE CPU and a 2560-core Blackwell GPU with 128GB LPDDR5X; the T4000 gives 1200 FP4 TFLOPS with 64GB. The QSFP28 cage runs at 100Gbps on T5000 or 75Gbps on T4000, alongside one 5GbE port. Display is one HDMI 2.0 and one DisplayPort 1.4 output, 16 MIPI CSI-2 lanes take cameras or capture cards, and a T5000 build can encode up to six 4K60 H.265 streams.

Four USB 3.2 Gen2 Type-A and two Type-C ports, line in and out, a PCIe Gen5 x4 gold finger and M.2 E, M and B key slots cover peripherals and expansion. A 40-pin header carries UART, GPIO, I2C, I2S, SPI and PWM, and T5000 builds add CAN bus. Yuan’s QCAP SDK and optional Xtreamer software cover RTSP, RTMP, HLS, SRT, TS and WebRTC streaming, with NDI-HX and Full NDI needing a separate licence. The EDG6N0-S T5X MB is the same board without the enclosure, and the EDG6N0-S T5X NVR adds eight PoE camera ports and two SATA drive bays.

The EDG6N0-S T5X runs Ubuntu 24.04.2 with JetPack 7.0.0, takes 20 to 57V DC, operates from -20 to 45°C with airflow and carries CE and FCC marks. We supply it configured and tested, with the Jetson module, NVMe storage and JetPack image fitted before it ships.