Quick Specs: Jetson Orin Nano Super | Jetson AGX Orin | Jetson Thor
Overview
NVIDIA Jetson is the agentic-ready AI platform for students, educators, and researchers building real-world AI. Professional-grade compute, open generative AI model support, and a global developer community — everything you need to go from idea to something real.
Features
Jetson ships with agent skills and open generative AI model support—no setup gymnastics required.
From autonomous machines to classroom bots, Jetson is purpose-built for AI that operates in the physical world.
Start with Jetson Orin Nano Super, scale to Jetson AGX Thor. One platform grows with your ambitions.
Join a global community of Jetson developers. Get answers, share projects, and learn from builders at every level.
Educational Benefits
Ready to build? Find your NVIDIA Jetson Developer Kit — from the Jetson Orin Nano™ Super to Jetson Thor™ — and buy today through a trusted partner.
Your first robot deserves a real brain. Jetson Orin Nano Super is compact enough for your backpack and powerful enough to run real AI right out of the box. Build with Python, explore agentic skills, and get hands-on with the same technology powering real-world autonomous machines. No NVIDIA CUDA™ mastery required. Just build.
Where classroom projects become real-world AI. Jetson AGX Orin brings serious compute to your research and coursework, handling computer vision, generative AI, and autonomous navigation with room to spare. Faculty rely on it across curriculum levels. Researchers depend on it for demanding workloads. Whatever you're building next, Jetson AGX Orin is ready.
This is the brain your robot has been waiting for. The most powerful platform in the Jetson lineup, Jetson Thor is the compute backbone for next-generation humanoids, multi-robot fleets, and autonomous systems that push the field forward. Run the largest reasoning models locally in real time, deploy complex vision-language AI at the edge, and build what others are still theorizing.
Specifications
Every student's project is different. Compare compute, memory, power, and more to find the NVIDIA Jetson developer kit that's right for your coursework, research, or lab.
| | NVIDIA Jetson Orin Nano Super<br /> Developer Kit | NVIDIA Jetson AGX Orin<br /> Developer Kit | NVIDIA Jetson AGX Thor<br /> Developer Kit |
|---|---|---|---|
| AI Performance | 67 TOPS | 275 TOPS | 2070 TFLOPS (FP4—sparse) |
| Module | Jetson Orin Nano 8GB | Jetson AGX Orin 64GB | Jetson T5000 |
| GPU | 1024-core NVIDIA Ampere architecture GPU<br /> with 32 Tensor Cores | 2048-core NVIDIA Ampere architecture GPU<br /> with 64 Tensor Cores | 2560-core NVIDIA Blackwell architecture GPU with fifth-gen Tensor Cores<br /> Multi-Instance GPU (MIG) with 10 TPCs |
| CPU | 6-core Arm® Cortex®-A78AE v8.2 64-bit CPU<br /> 1.5 MB L2 + 4 MB L3 | 12-core Arm® Cortex®-A78AE v8.2 64-bit CPU<br /> 3 MB L2 + 6 MB L3 | 14-core Arm® Neoverse®-V3AE 64-bit CPU<br /> 1 MB L2 Cache per core<br /> 16 MB Shared System L3 Cache |
| Memory | 8 GB 128-bit LPDDR5<br /> 102 GB/s | 64 GB 256-bit LPDDR5<br /> 204.8 GB/s | 128 GB 256-bit LPDDR5X<br /> 273 GB/s |
| Networking* | 1x GbE | RJ45 connector with up to 10GbE | 1x 5GbE RJ45 connector 1x QSFP28 (4x 25GbE) |
| I/O | 4x USB-A 3.2 Gen2<br /> USB Type-C (UFP)<br /> 1x DisplayPort 1.2 (+MST)<br /> 40-pin expansion header (UART, SPI, I2S, I2C, GPIO)<br /> 12-pin button header | 4-pin fan header<br /> microSD slot | DC power jack | USB Type-C: 2x USB 3.2 Gen2<br /> USB Type-A: 2x USB 3.2 Gen2, 2x USB 3.2 Gen1<br /> USB Micro-B: USB 2.0<br /> 1x DisplayPort 1.4a (+MST)<br /> 40-pin header (UART, SPI, I2S, I2C, CAN, PWM, DMIC, GPIO)<br /> 12-pin automation header | microSD slot | DC power jack | QSFP connector<br /> HDMI port<br /> DisplayPort<br /> 2x USB-A 3.2 | 2x USB-C 3.1<br /> Gigabit Ethernet<br /> 2x 13-pin CAN header<br /> Microfit power jack |
| Storage | SD Card Slot + external NVMe via M.2 Key M | 64 GB eMMC 5.1 | 1 TB NVMe M.2 Key M slot |
| Power | 7 W-25 W | 15 W-60 W | 40 W-130 W |
* Refer to the Software Features section of the latest NVIDIA Jetson Linux Developer Guide for a list of supported features.
Resources
Everything you need to learn AI and robotics, build your first project, and connect with developers around the world.
Jetson is an agentic-ready AI platform designed to help students, educators, and researchers build real-world robotics and autonomous systems. Jetson provides professional-grade compute power that allows you to move from your first simple project to complex, industry-standard robotics applications.
Explore the Jetson AI Lab for tutorials and projects to see what you can build.
Agentic skills refer to pre-built capabilities and generative AI support that allow your robots to perceive, reason, and act autonomously without requiring you to write every line of code from scratch. This allows you to focus on the unique aspects of your project rather than low-level setup.
Review the JetPack components and agentic software to understand the software stack.
These kits represent different levels of compute power to grow with your ambitions. The Jetson Orin Nano Super is the ideal entry point for beginners and students, the Jetson AGX Orin offers more power for coursework and research, and Jetson Thor is the most powerful platform, designed for advanced robotics, humanoid research, and multi-robot fleets.
Compare the Jetson Developer Kits to choose the best one for your needs.
No. The Jetson platform is designed to be accessible, allowing you to build with familiar languages like Python and utilize existing agentic skills and open generative AI models right out of the box.
Learn how to start your first project with Jetson Getting Started Guides.
While Jetson is purpose-built for robotics and autonomous machines, it’s also highly effective for computer vision, data analytics, and running large language models (LLMs). It serves as a comprehensive desktop-based AI platform for any computationally complex work.
View examples of edge applications and AI frameworks you can develop.
The Jetson AI Lab is a central resource for developers to explore optimized tutorials, benchmarks, and generative AI models that are specifically designed to run on the Jetson platform. It is the best place to find project inspiration and step-by-step guidance.
Browse projects and tutorials at the Jetson AI Lab portal.
You can connect with a global community of makers, students, and professional developers on the Jetson Forums. It is a great place to ask questions, troubleshoot, and share your own projects with others.
Join the conversation on the Jetson Developer Forums.
Yes, Jetson Developer Kits are fully compatible with major AI frameworks, running efficiently on the Arm architecture. You can use the same libraries used in industry to develop and test your models locally.
Explore the software libraries and AI frameworks supported by JetPack.
Absolutely. You can integrate your Jetson kit with NVIDIA tools like Isaac for robotics simulation, Metropolis for smart city applications, and Holoscan for real-time sensor processing, helping you build more advanced edge applications.
Learn more about developing with NVIDIA AI frameworks to expand your project capabilities.
Jetson is an "agentic-ready" platform that combines hardware and software optimized for the physical world. It provides the performance needed for real-time autonomous navigation, vision, and reasoning, whereas standard PCs often lack the dedicated hardware acceleration needed for mobile robotics.
See why educators rely on Jetson for hands-on experience in their curriculum.
Yes, the NVIDIA Deep Learning Institute (DLI) provides hands-on courses and workshops designed for students, educators, and researchers at every level. These courses are excellent for gaining foundational and advanced skills in AI and robotics.
Start your training journey at the NVIDIA Deep Learning Institute.
Yes, the Jetson platform is designed to grow with your ambitions. Because the kits share the same software stack, you can prototype on a Jetson Orin Nano Super and later scale up your project to a more powerful kit like the Jetson AGX Orin or Jetson Thor without needing to relearn the entire platform.
Compare the full lineup in the Jetson Developer Kits specifications.
Yes, the Jetson AGX Orin and Jetson Thor are specifically positioned as the compute backbone for advanced research. They are capable of handling demanding workloads like computer vision, generative AI, and autonomous navigation, making them ideal for high-level coursework and prototypes.
Read more about Jetson for Researchers and Educators on our dedicated resources page.
Jetson Developer Kits generally require you to flash the JetPack SDK, which includes the operating system, libraries, and tools necessary for development. The "Getting Started" guides provide simple instructions on how to set this up on your first day.
Follow the Jetson Getting Started Guide for setup instructions.
You can purchase Jetson Developer Kits from a variety of authorized NVIDIA partners. The campaign page provides a list of trusted partners to help you find and buy the kit that fits your needs.
Locate a partner and buy your kit through our partner network.
Yes, the Jetson community is very active in sharing their robotics and AI projects. You can visit the Jetson AI Lab and the forums to see what other students and researchers have built, which can help provide ideas and technical guidance for your own work.
Get inspired by community-built projects.
Power requirements vary by kit, with the smaller kits being designed for compact, battery-powered robotics and the larger kits requiring more power for advanced compute tasks. Check the individual datasheet for your specific kit to ensure you have the correct power supply.
Access technical specifications and power requirements in our comparison table.
Yes, Jetson kits are designed to run open generative AI models locally, which is crucial for building autonomous agents that need to process information without relying on the cloud, ensuring lower latency and better privacy.
Find supported models in the Jetson AI Lab tutorials.