▶ NVIDIA® Jetson AGX Thor (T4000 / T5000) Module
Flagship Blackwell Architecture, Ultimate On-Device AI Performance
▶ Up to 2070 Sparse FP4 TFLOPs AI Computing Power
Smooth Local Inference for 7B–70B Large Language Models
▶ Up to 14-core Arm® Neoverse®-V3AE 64-bit CPU
Strong Multi-tasking & Real-time Processing Capability
▶ 128 GB High-Bandwidth LPDDR5X (273 GB/s)
Alleviate Large Model Memory Bottleneck
▶ 2 × 10GbE Ethernet + Wi-Fi 6 + BT 5.3
Ultra-High-Speed Data Transmission & Stable Wireless Connection
▶ 2 × HDMI® 2.1 TMDS (4K@60 Hz) + Rich High-Speed I/O
4 × USB 3.2 Gen 1 Type-A + 1 × USB 3.2 Gen 2 Type-C
▶ 1 TB NVMe SSD
Large Storage for Models, Datasets & Applications
▶ Compact Desktop Design
130 × 125 × 75 mm, Plug-and-Play, No Rack Required
▶ Local Private Deployment
All Data & Models Processed On-premises, No Cloud Upload Needed
▶ Complete NVIDIA AI Ecosystem
JetPack 7.1, CUDA, TensorRT, NIM, Container Support