Contact us
中文
Coeus - 4000G8
NVIDIA® Jetson Thor™ T4000
1200 TFLOPS AI Compute
Ideal for general humanoid robots, autonomous driving and generative AI processing
Compatible with multiple AI frameworks
Expandable with 8-channel GMSL, 5G, 4G & Wi-Fi interfaces
- Introduction
- Specification
- Order
- Download
Powered by NVIDIA’s latest-generation Jetson Thor T4000 module, this device delivers up to 1200 TFLOPS AI Compute, delivering high-performance hardware dedicated to humanoid robots, autonomous driving, spatial intelligence, generative AI and other scenarios.
Boasting rich expandability, it supports expansions including 8-channel GMSL cameras, 5G, 4G and Wi-Fi. Multi-mode communication with terminal devices eliminates redundant intermediate data forwarding and processing, drastically cutting communication latency while protecting data privacy. Comprehensive expansion options greatly simplify deployment workflows for users.
Its Blackwell-architecture GPU efficiently runs large Transformer models, delivering robust computing power for generative AI and real-time edge inference. NVIDIA’s full-stack software ecosystem built for edge AI and robotics covers every stage from model training to field deployment.
| CPU | 12-core Arm® Neoverse®-V3AE 64-bit 1MB L2 +16MB L3 2.6 GHz |
| GPU | 1536-core NVIDIA Blackwell GPU 1.53GHz |
| Vision accelerator | 1 × PVA v3 |
| RAM | 64 GB 256-bit LPDDR5X |
| Storage | NVMe SSD support via PCIe |
| AI Compute | 2070 TFLOPS (FP4 — Sparse) |
| I/O Interface-Front | 1 × Power Switch 3 × RJ45 2.5G LAN 4 × USB 3.2 1 × COM (Not full signals, 1×DB9, RS-232/422/485 configurable) 1 × GPIO（8DI & 8DO） 4 × Indicator （PWR、SYS、GPO1、GPO2） |
| I/O Interface-Bottom | 1 × Recovery 1 × OTG（Type-C） 1 × 3P Power Connector 1 × HDMI 1 × Line Out 1 × Nano-SIM 6 × Antenna holes 1 × Ground screw hole |
| I/O Interface-Top | 8 × GMSL |
| Expand | 1 × M.2 E-Key 2230 (supports PCIe x1/USB2.0 signal) 1 × M.2 M-Key 2280 (supports PCIe x4 signal) 1 × M.2 B-Key 3052 (supports USB3.0/USB2.0/SIM signal) |
| Cooling Method | 2 × DC 12V 7015 |
| Power | DC 12-30V |
| Dimension | 96.8mm (W) x 137mm (D) x 226mm (H)(with stand) |
| Installation Method | all-mounted, rail-mounted |
| Operating Temperature | -10 ~ 50 ℃ with air flow |
| Storage Temperature | -20 ~ 80 ℃ |
| Working Humidity | 40℃，95% Relative humidity, non-condensing |
| Name | Coeus-4000G8 |
| AI Compute | 1200 TFLOPS (FP4 — Sparse) |
| RAM | 64 GB |
| Order Number | 10JD4000T02XH |
| Core | Jetson Thor T4000 64GB |
Order or get information, please send an email to: sales@nexgemo.cn
