NVIDIA Jetson Orin Nano: Specs, Setup, and First Projects
The NVIDIA Jetson Orin Nano is the most accessible way to run serious AI at the edge. With a 1024-core Ampere GPU, up to 67 TOPS of AI performance after the "Super" update, and a configurable power envelope that tops out around 25 W, it sits in a sweet spot between hobbyist boards like the Raspberry Pi and full workstation GPUs. This guide covers what the Orin Nano actually is, the specs that matter, how to set it up with JetPack step by step, how it compares honestly with a Raspberry Pi 5 plus AI HAT, and the first projects worth building on it.
Disclosure: this article contains affiliate links. If you buy through them we may earn a small commission at no extra cost to you. We only recommend hardware we would use ourselves.
What is the Jetson Orin Nano?
The Jetson Orin Nano is a compact system-on-module (SoM) from NVIDIA's Jetson embedded computing line, aimed at edge AI: computer vision, robotics, and — increasingly — small generative AI models running locally. Most people buy it as the Jetson Orin Nano Developer Kit, which pairs the 8 GB Orin Nano module with a reference carrier board that exposes the ports you need to start building: DisplayPort, USB, Gigabit Ethernet, two MIPI CSI camera connectors, a 40-pin GPIO header, and two M.2 slots (one for an NVMe SSD, one pre-populated with a Wi-Fi module).
Unlike a Raspberry Pi, which is a general-purpose single-board computer, the Orin Nano is an AI accelerator first and a desktop-class Linux machine second. Its GPU runs the same CUDA and TensorRT software stack as NVIDIA's data-center hardware, which means models and pipelines developed on a big GPU can usually be optimized down to the Jetson with far less friction than porting to any other edge platform.
The "Super" revision: same hardware, big performance jump
In late 2024 NVIDIA rebranded the kit as the Jetson Orin Nano Super Developer Kit and cut its price roughly in half. The interesting part: the "Super" boost is mostly a software and firmware unlock. A new maximum power mode raises GPU and memory clocks, lifting peak AI performance from 40 TOPS to 67 TOPS — about a 1.7× jump — and increasing memory bandwidth from 68 GB/s to 102 GB/s. Existing owners of the original Orin Nano Developer Kit get the same uplift free of charge by updating firmware and installing JetPack 6.2 or later. If you are buying today, everything on sale is effectively the Super configuration; just make sure you flash a current JetPack so you are not leaving a third of the performance on the table.
- The NVIDIA Jetson Orin Nano Developer Kit sets a new standard for creating entry-level AI-powered robots, smart drones, and intelligent cameras,and simplifies getting started with the Jetson Orin Nano series. Compact design, lots of connectors and up to 40 TOPS of AI performance make this developer kit perfect for transforming your visionary concepts into reality. With up to 80X the performance of Jetson Nano, it can run all modern AI models, including transformer and advanced robotics models.
- The developer kit comprises a Jetson Orin Nano 8GB module and a reference carrier board that can accommodate all Orin Nano and Orin NX modules, providing an ideal platform for prototyping your next-gen edge AI product. The Jetson Orin Nano 8GB module features an Ampere GPU and a 6-core ARM CPU, enabling multiple concurrent AI application pipelines and high-performance inference. The carrier board boasts a wide array of connectors, including two MIPI CSI connectors supporting camera modules with up to 4-lanes, allowing higher resolution and frame rate than before.
- Jetson runs the NVIDIA AI software stack, with available use-case-specific application frameworks, including NVIDIA Isaac for robotics, DeepStream for vision AI, and Riva for conversational AI. You can save significant time with NVIDIA Omniverse Replicator for synthetic data generation (SDG), and with NVIDIA TAO Toolkit for fine-tuning pretrained AI models from the NGC catalog.
- Ecosystem partners offer additional AI and system software, developer tools, and custom software development. They can also help with cameras and other sensors, as well as carrier boards and design services for your product.
- Jetson Orin modules are unmatched in performance and efficiency for robots and other autonomous machines, and give you the flexibility to create the next generation of AI solutions with the latest NVIDIA technology. Together with the world-standard NVIDIA AI software stack and an ecosystem of services and products, your road to market has never been faster.
- 【Core Parameters】★AI Perf:34-67 TOPS ★GPU:512-core NVIDIA Ampere architecture GPU with 16 Tensor Cores ★CPU:6-core Arm Corte-A78AE v8.2 64-bit CPU 1.5MB L2 + 4MB L3 ★Memory:4GB 64-bit LPDDR5 51 GB/s ★Storage: external NVMe via M.2 Key M (NOTE:SUB Board No SD Card Slot)
- 【Empowered by Large Al Model, Enhanced Human-Computer Interaction】Jetson Orin Super leverages three AI models and incorporates an AI voice interaction module. This multimodal visual system matches the scene being described, enabling environmental awareness and AI visual gameplay. Combined with a large-scale voice module and camera, it enables speech-to-text, semantic analysis, natural conversation, and real-time video analysis, enabling advanced embodied AI applications.
- 【AI Upgrade】Jetson Orin Nano series modules are compact in size but can deliver up to 34-67 TOPS of AI performance, with power consumption ranging from 7 watts to 25 watts. Compared to the Jetson Nano B01, it offers up to 80 times the performance and sets a new standard for entry-level edge AI.
- 【Highly compatible carrier board】Yahboom's carrier board is fully compatible with orin nano module. Compared to carrier boards that use Jetson Nano on the market, the newly upgraded circuit supports 25W power mode, which enables larger and more complex neural networks and fully leverages the performance of the core module. The resources, size, and interfaces of the Yahboom carrier board are consistent with the official board, with the only difference addition of power switch button.
- 【Tutorial materials provided】The JETSON system based on Ubuntu 22.04 provides a complete desktop Linux environment with accelerated graphics, supporting NVIDI-ACUDA 12.6, TensorRT 10.7.0, cuDNN 9.6.0, OpenCV 4.10.0, etc. The performance on AI LLM, VLM and visual Transformer is significantly improved compared with the previous generation.
- The NVIDIA Jetson AGX Orin 64GB Developer Kit makes it easy to get started with Jetson Orin. Compact size, lots of connectors, and up to 275 TOPS of AI performance make this developer kit perfect for prototyping advanced AI-powered robots and other autonomous machines.
- The developer kit includes a Jetson AGX Orin 64GB module, and can emulate all the Jetson Orin modules. It supports multiple concurrent AI application pipelines with the NVIDIA Ampere GPU architecture, next-generation deep learning and vision accelerators, high-speed IO and fast memory bandwidth. Now you can develop solutions using your largest and most complex AI models to solve problems such as natural language understanding, 3D perception, and multi-sensor fusion.
- Jetson runs the NVIDIA AI software stack, and use-case specific application frameworks are available, including Isaac for robotics, DeepStream for vision AI, and Riva for conversational AI. You can save significant time with NVIDIA Omniverse Replicator for synthetic data generation (SDG), and by using NVIDIA TAO toolkit to fine-tune pretrained AI models from the NGC catalog.
- Jetson ecosystem partners offer additional AI and system software, developer tools, and custom software development. They can also help with cameras and other sensors, as well as carrier boards and design services for your product.
- With the computing capability of more than 8 Jetson AGX Xavier systems in a developer kit that integrates the latest NVIDIA GPU technology with the world’s most advanced deep learning software stack, you’ll have the flexibility to create tomorrow’s AI solution as well as today’s.
Key specs that actually matter
Spec sheets for embedded boards are long; these are the numbers that change what you can build:
- GPU: 1024-core NVIDIA Ampere architecture GPU with 32 Tensor Cores. This is the same architecture family as the RTX 30 series, so it supports CUDA, cuDNN and TensorRT natively.
- AI performance: up to 67 TOPS (INT8, sparse) in the Super power mode; 40 TOPS in the original configuration.
- CPU: 6-core Arm Cortex-A78AE at up to 1.7 GHz — comfortable for preprocessing, ROS nodes and general Linux use.
- Memory: 8 GB LPDDR5, shared between CPU and GPU, with 102 GB/s bandwidth in Super mode. Shared memory is a real advantage for LLM workloads: the GPU can address almost all of it.
- Storage: microSD slot plus an M.2 Key M slot for NVMe SSD. An SSD is strongly recommended — it is dramatically faster and more reliable than SD for AI workloads.
- Power: configurable modes from 7 W to 25 W (the Super MAXN mode). You choose the trade-off between thermals, power draw and throughput with a single command (
nvpmodel). - I/O: 2× MIPI CSI camera connectors, 4× USB 3.2, DisplayPort 1.2, Gigabit Ethernet, 40-pin header (GPIO, I2C, SPI, UART) for sensors and robotics.
The practical takeaway: with 8 GB of fast shared memory and Ampere Tensor Cores, the Orin Nano can run real-time object detection at high frame rates, several vision models concurrently, or quantized language models in the 3–8 B parameter class — things that are out of reach for CPU-only boards.
Setting up the Jetson Orin Nano step by step
Setup is a little more involved than a Raspberry Pi, mainly because of one firmware gotcha. Here is the reliable path.
1. Gather what you need
- The developer kit and its power supply.
- A microSD card of 64 GB or more (UHS-1), or better, an NVMe SSD for the M.2 slot.
- Monitor (DisplayPort), keyboard, mouse, and a network connection.
- A separate computer to flash the card image (any OS works for the SD route).
2. Check the firmware version first
Kits that sat in a warehouse may ship with factory firmware that predates JetPack 6 and will not boot a JetPack 6.x SD card. If your board's UEFI firmware is older than version 36.x, you need to update it first — either by booting an older JetPack 5.1.3 image once and letting it update, or by flashing from an Ubuntu host with NVIDIA SDK Manager. Recently manufactured Super kits generally ship ready for JetPack 6, but checking the firmware version on the boot screen saves a confusing "black screen" afternoon.
3. Flash the JetPack image
Download the latest JetPack SD card image (6.2 or newer to get the Super performance modes) from NVIDIA's JetPack page, and write it to the microSD card with Balena Etcher. If you prefer the SSD-only route or need to recover a board, use NVIDIA SDK Manager on an Ubuntu 20.04/22.04 host connected via USB-C with the board in recovery mode — more steps, but it flashes firmware, bootloader and OS in one pass.
4. First boot and configuration
Insert the card in the slot on the underside of the module, connect peripherals, and power on. The first boot walks you through the usual Ubuntu-style setup: accept the EULA, pick language and time zone, create a user, and connect to the network. Once on the desktop, two commands matter:
sudo apt update && sudo apt upgrade— pulls current JetPack components.sudo nvpmodel -m 2(or select MAXN SUPER in the desktop power menu) — enables the 67 TOPS power mode. Add a decent airflow path; the bundled fan handles it, but don't smother the board in a closed drawer.
5. Verify the AI stack
Run jtop (install with sudo pip3 install jetson-stats) to confirm GPU clocks and the active power mode. Then pull a container from NVIDIA's jetson-containers project — for example an Ollama or a YOLO TensorRT image — and run a first inference. If a quantized small LLM answers you from the board itself, everything is wired correctly.
Jetson Orin Nano vs Raspberry Pi 5 + AI HAT: an honest comparison
This is the question most buyers are really asking, and the honest answer is that they are different tools.
A Raspberry Pi 5 with the AI HAT+ (Hailo accelerator) costs roughly half as much all-in, sips power, and is excellent at the specific thing the Hailo chip does: running compiled vision models (detection, pose, segmentation) at impressive frame rates. Its ecosystem of HATs, cameras and tutorials is unmatched — we covered that side in our guide to the Raspberry Pi AI Camera: setup, specs and first projects, and if you lean that way, our roundup of the best Raspberry Pi AI kits compares the current options.
The Orin Nano's advantage is generality and headroom. The Hailo accelerator only runs models compiled for it, and the Pi's memory is not addressable by the accelerator, which rules out local LLMs of any useful size. The Jetson runs the full CUDA ecosystem: PyTorch, TensorRT, DeepStream, Isaac ROS, plus quantized 3–8 B LLMs and vision-language models in its 8 GB of shared memory. When your project mixes several models, needs custom architectures, or will ever graduate to production edge hardware, the Jetson path is smoother.
Choose the Pi 5 + AI HAT if: you want the cheapest capable vision setup, you're running one standard detection model, battery/solar power matters, or you're already deep in the Pi ecosystem.
Choose the Jetson Orin Nano if: you want local LLMs or multimodal models, multi-camera or multi-model pipelines, robotics with ROS 2 and hardware acceleration, or a learning platform whose skills (CUDA, TensorRT) transfer directly to industry hardware.
First projects worth building
Once the board is set up, these three project lanes give the best return on your first weekends.
Real-time computer vision
Start with YOLO (v8 or newer) exported to TensorRT: a CSI or USB camera, one container, and you have real-time object detection at high FPS with the GPU barely warm. From there, DeepStream lets you chain decoding, inference and tracking for multi-stream setups — the same pattern used in commercial video analytics. A natural next step is fusing the camera with other sensors (IMU, LiDAR, radar) for more robust perception; our pillar guide on what sensor fusion is and how it works explains the concepts you'll be implementing.
Small local LLMs and vision-language models
This is what the Super update was marketed on, and it delivers within reason. Quantized models in the 2–8 B range (Llama 3.2 3B, Qwen 2.5, Phi-3, or vision-language models like LLaVA variants) run locally via Ollama or NVIDIA's jetson-containers at usable interactive speeds. Realistic expectations: a 3 B model feels snappy, an 8 B model is workable, anything larger is not what this board is for. A private, always-on home assistant that never sends audio to the cloud is a very achievable first build.
Robotics with ROS 2
The Orin Nano is the entry ticket to NVIDIA's Isaac ROS stack: hardware-accelerated packages for visual SLAM, depth estimation, AprilTag detection and manipulation. Pair the board with a wheeled chassis, a depth camera and the 40-pin header for motor control, and you have a genuine autonomous-navigation testbed — the same software patterns used on much larger robots.
- The NVIDIA Jetson Orin Nano Developer Kit sets a new standard for creating entry-level AI-powered robots, smart drones, and intelligent cameras,and simplifies getting started with the Jetson Orin Nano series. Compact design, lots of connectors and up to 40 TOPS of AI performance make this developer kit perfect for transforming your visionary concepts into reality. With up to 80X the performance of Jetson Nano, it can run all modern AI models, including transformer and advanced robotics models.
- The developer kit comprises a Jetson Orin Nano 8GB module and a reference carrier board that can accommodate all Orin Nano and Orin NX modules, providing an ideal platform for prototyping your next-gen edge AI product. The Jetson Orin Nano 8GB module features an Ampere GPU and a 6-core ARM CPU, enabling multiple concurrent AI application pipelines and high-performance inference. The carrier board boasts a wide array of connectors, including two MIPI CSI connectors supporting camera modules with up to 4-lanes, allowing higher resolution and frame rate than before.
- Jetson runs the NVIDIA AI software stack, with available use-case-specific application frameworks, including NVIDIA Isaac for robotics, DeepStream for vision AI, and Riva for conversational AI. You can save significant time with NVIDIA Omniverse Replicator for synthetic data generation (SDG), and with NVIDIA TAO Toolkit for fine-tuning pretrained AI models from the NGC catalog.
- Ecosystem partners offer additional AI and system software, developer tools, and custom software development. They can also help with cameras and other sensors, as well as carrier boards and design services for your product.
- Jetson Orin modules are unmatched in performance and efficiency for robots and other autonomous machines, and give you the flexibility to create the next generation of AI solutions with the latest NVIDIA technology. Together with the world-standard NVIDIA AI software stack and an ecosystem of services and products, your road to market has never been faster.
- 【Core Parameters】★AI Perf: 34/67 TOPS ★GPU:1024-core official Ampere architecture GPU with 32 Tensor Cores ★CPU:6-core Arm Corte-A78AE v8.2 64-bit CPU 1.5MB L2 + 4MB L3 ★Memory:8GB 128-bit LPDDR5 68 GB/s ★Storage: external NVMe via M.2 Key M
- 【Empowered by Large Al Model, Enhanced Human-Computer Interaction】Jetson Orin Super leverages three AI models and incorporates an AI voice interaction module. This multimodal visual system matches the scene being described, enabling environmental awareness and AI visual gameplay. Combined with a large-scale voice module and camera, it enables speech-to-text, semantic analysis, natural conversation, and real-time video analysis, enabling advanced embodied AI applications.
- 【AI Upgrade】Jetson Orin Nano series modules are compact in size but can deliver up to 34-67 TOPS of AI performance, with power consumption ranging from 7 watts to 25 watts. Compared to the Jetson Nano B01, it offers up to 80 times the performance and sets a new standard for entry-level edge AI.
- 【Highly compatible carrier board】Yahboom's carrier board is fully compatible with orin nano module. Compared to carrier boards that use Jetson Nano on the market, the newly upgraded circuit supports 25W power mode, which enables larger and more complex neural networks and fully leverages the performance of the core module. The resources, size, and interfaces of the Yahboom carrier board are consistent with the official board, with the only difference addition of power switch button.
- 【Tutorial materials provided】The JETSON system based on Ubuntu 22.04 provides a complete desktop Linux environment with accelerated graphics, supporting CUDA 12.6, TensorRT 10.7.0, cuDNN 9.6.0, OpenCV 4.10.0, etc. The performance on AI LLM, VLM and visual Transformer is significantly improved compared with the previous generation.
Frequently asked questions about the Jetson Orin Nano
Is the Jetson Orin Nano Super a different board from the Orin Nano?
No. The Super Developer Kit is the same hardware with new firmware and JetPack software that unlock a higher power mode: 67 TOPS instead of 40, and 102 GB/s of memory bandwidth instead of 68. Original kits get the same boost with a free firmware and JetPack 6.2+ update.
Can the Jetson Orin Nano run large language models?
It runs small LLMs well. Quantized models up to roughly 8 B parameters fit in its 8 GB of shared memory and run at interactive speeds — good for local assistants, summarization and RAG experiments. It is not a platform for 13 B+ models.
Do I need an SSD, or is a microSD card enough?
A 64 GB+ UHS-1 microSD card is enough to get started, and it's the simplest flashing route. For real work, add an NVMe SSD in the M.2 slot: models and containers are large, and the SSD is far faster and more durable.
How much power does the Jetson Orin Nano use?
It is configurable between 7 W and 25 W. The full 67 TOPS requires the MAXN "Super" mode at the top of that range; lower modes trade throughput for cooler, quieter, battery-friendly operation.
Next steps
The Jetson Orin Nano is the rare dev board that stays useful after the honeymoon: it starts as a learning platform and scales into real perception and robotics work. If you're still weighing platforms, read our hands-on guide to the Raspberry Pi AI Camera and the comparison of the best Raspberry Pi AI kits for the budget side of the decision, then browse the rest of our Edge AI guides for what to build next.
Last update 2026-10-03. Price and product availability may change.
Recommended: