NVIDIA Jetson T5000 Module

The NVIDIA® Jetson T5000™ sets new standards in performance for Physical AI and robotics. With up to 2070 FP4 TFLOPS of AI performance and 128 GB of memory, it delivers massive computing power – all with flexible power settings ranging from 40 to 130 W. Compared to the AGX Orin™, it offers over 7.5 times more AI performance while achieving 3.5 times better energy efficiency.

At the heart of the module is an NVIDIA Blackwell GPU with Transformer Engine and Multi-Instance GPU (MIG) support, capable of effortlessly handling even the latest generative AI models. Jetson Thor accelerates real-time multi-sensor applications with a 14-core Arm® Neoverse®-V3AE CPU, 4x 25 GbE network interfaces, and a wide range of I/O options for sensor fusion. Powerful accelerators such as the third-generation Programmable Vision Accelerator (PVA), dual encoders/decoders, an optical flow accelerator, and more are also included.

Jetson Thor belongs to a new generation of robot computers designed from the ground up for next-generation Physical AI. It supports a wide range of generative AI models – from Vision Language Action (VLA) models like NVIDIA Isaac™ GR00T for humanoids to all common LLMs and VLMs. For a seamless cloud-to-edge experience, Jetson Thor runs on the NVIDIA AI software stack, including NVIDIA Isaac for robotics, NVIDIA Metropolis for visual AI, and NVIDIA Holoscan for sensor processing. Edge AI agents can also be easily created – for example, using workflows like Video Search and Summarization (VSS).

Our partner ecosystem provides everything you need for your projects: carrier boards, design services, cameras, sensors, and additional AI and system software. This enables faster development of solutions in robotics, smart spaces, retail, industry, healthcare, and many other fields.

With the Jetson Thor module, you get production-ready performance, enormous AI power, and extensive sensor capabilities – all in a compact, high-performance package. Ideal for developers looking to take humanoid robotics or other Physical AI applications to the next level.

Technical Specifications

AI Performance

• 2070 TFLOPS (FP4–sparse)

GPU

• 2560-core NVIDIA Blackwell architecture
• GPU with 96 fifth-generation Tensor Cores
• Multi-Instance GPU with 10 TPCs
• Maximum GPU frequency: 1.57 GHz

CPU

• 14-core Arm® Neoverse®-V3AE 64-bit CPU
• 64 KB I-Cache, 64 KB D-Cache
• 1 MB L2 cache per core
• 16 MB shared system L3 cache
• Maximum CPU frequency: 2.6 GHz

Vision Accelerator

• 1x PVA v3

Memory

• 128 GB 256-bit LPDDR5X
• 273 GB/s

Storage

• Supports NVMe via PCIe
• Supports SSD via USB3.2

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)

Video Decode

• 4x 8Kp30 (H.265)
• 10x 4Kp60 (H.265)
• 22x 4Kp30 (H.265)
• 46x 108