Jetson Orin Nano vs Orin NX vs AGX Orin

Pick the Jetson that fits your cameras. Roboflow runs on all of them.

Orin Nano, Orin NX, and AGX Orin share NVIDIA's Orin architecture but range from up to 67 to up to 275 TOPS. Train a vision model in Roboflow, deploy it with Roboflow Inference on whichever Jetson fits your cameras and budget, and move to a bigger module as you grow without retraining. Over 2 million developers build with Roboflow.

An automated machine vision inspection system on a conveyor belt detecting a torn flap defect on a cardboard box.

Why are companies choosing Roboflow?

Roboflow is enterprise vision AI infrastructure, used by more than half of the Fortune 100, with SOC 2 Type II, HIPAA, and air-gapped deployment. Create datasets, train models, and deploy to production.

  • One Model, Any Jetson

    Train once in Roboflow and deploy the same model to Orin Nano, Orin NX, or AGX Orin with one install command.

  • Hardware Ready to Ship

    Roboflow ships Jetson-based devices, from the AI1 on Orin NX 8GB to AGX Orin 64GB, so you can skip sourcing and setup.

  • Runs Next to Your Cameras

    Process video on the Jetson itself, with TensorRT acceleration when you need faster models.

Which Jetson fits your cameras?

All three run the same Roboflow model, so you can start small and move up without retraining.

  • Jetson Orin Nano: Start Small

    Up to 67 TOPS at 7W to 25W. Best for prototypes and one or two cameras running a lightweight model.

  • Jetson Orin NX: Most Production Lines

    Up to 157 TOPS in the same footprint as Orin Nano. Best for a few cameras and real-time detection on the line.

  • Jetson AGX Orin: Many Streams

    Up to 275 TOPS and up to 64GB of memory. Best for many cameras, larger models, or several models at once.

OK I’m ready, how do I use Roboflow?

Roboflow makes it easy to build, train, and deploy custom computer vision models, even if you’re not a machine learning expert.

Talk to a Vision AI engineer before you pick a Jetson

Not sure which Jetson your cameras need? Bring Roboflow your use case and we'll size the hardware and show you a model running on it on the first call.

Ask us about:

  • A recommendation on Orin Nano, Orin NX, or AGX Orin for your camera count
  • A live demo of a model trained on your own images
  • Roboflow hardware that ships with Jetson inside
  • Your trained model served over MCP, so frontier models can call it

Over 80,000 organizations build with Roboflow.

  • Rivian
  • Pella
  • Chobani
  • USG Corporation
  • BNSF Railway
  • American Woodmark
Start where you are

Compare Jetson Orin Nano, Orin NX and AGX Orin

Jetson Orin Nano, Orin NX, and AGX Orin all use NVIDIA's Orin architecture, but they're built for different workloads and budgets.

Orin Nano is the entry point, with up to 67 TOPS, 4GB or 8GB of memory, and 7W to 25W of power. Orin NX delivers up to 157 TOPS with 8GB or 16GB of memory in the same 69.6 x 45 mm footprint, so a carrier board built for Orin Nano also takes Orin NX. AGX Orin is the largest, with up to 275 TOPS, up to 64GB of memory, and twice the memory bandwidth of Orin NX, at 15W to 60W. Check current pricing before you budget a fleet.

If you need more than AGX Orin, Jetson Thor, NVIDIA's newest module, offers up to 2070 FP4 TFLOPS and 128GB of memory at 40W to 130W.

Whichever module you choose, Roboflow handles the vision model. Label images with Auto Label, train a model such as RF-DETR (opens in a new tab), Roboflow's open-source detection, segmentation, and keypoints model, and deploy it with Roboflow's open-source Inference server (opens in a new tab), which runs on JetPack 5 and 6 and picks the right container for your device with one install command. TensorRT acceleration is available when you need faster models, and the same model moves from Orin Nano to AGX Orin as your camera count grows.

Teams can also use Roboflow Workflows to build applications that combine computer vision models with business logic, then run them on NVIDIA Jetson next to your cameras, on-premises infrastructure, or the cloud.

For organizations building production computer vision applications in manufacturing, logistics, robotics, healthcare, or retail, Roboflow offers a complete training-to-deployment solution on the Jetson that fits each site.

Jetson Orin Nano vs Orin NX vs AGX Orin: Feature by Feature

FeatureJetson Orin NanoJetson Orin NXJetson AGX Orin
AI performanceUp to 67 TOPSUp to 157 TOPSUp to 275 TOPS
Memory4GB or 8GB LPDDR58GB or 16GB LPDDR532GB or 64GB LPDDR5
Memory bandwidth51 GB/s (4GB) or 102 GB/s (8GB)102.4 GB/s204.8 GB/s
Power7W to 25W10W to 40W15W to 60W
Size and connector69.6 x 45 mm, 260-pin SO-DIMM69.6 x 45 mm, 260-pin SO-DIMM (same footprint as Orin Nano)100 x 87 mm, 699-pin connector, integrated thermal transfer plate
Cameras (MIPI CSI-2)Up to 4 (8 with virtual channels)Up to 4 (8 with virtual channels)Up to 6 (16 with virtual channels)
Video decode (H.265)1x 4K60, or up to 11x 1080p301x 8K30, 2x 4K60, or up to 18x 1080p301x 8K30, 3x 4K60, or up to 22x 1080p30
Module price (1KU+, NVIDIA list)$349 (4GB), $399 (8GB); Orin Nano Super Developer Kit $399$649 (8GB), $999 (16GB)$1,799 (32GB), $2,999 (64GB); AGX Orin Developer Kit $3,499
Roboflow hardwareOrin Nano Super Developer Kit recommended as a starting pointRoboflow AI1 (Orin NX 8GB) and RF-NX16 (Orin NX 16GB)RF-AGX64 (AGX Orin 64GB)
Runs Roboflow modelsYes, Roboflow Inference on JetPack 5 and 6Yes, Roboflow Inference on JetPack 5 and 6Yes, Roboflow Inference on JetPack 5 and 6

Which One Should You Choose?

  • Choose Jetson Orin Nano to prototype, or for one or two cameras running a lightweight model where cost and power matter most.
  • Choose Jetson Orin NX for most production deployments: a few cameras, real-time detection, and a compact module that drops into the same carrier as Orin Nano.
  • Choose Jetson AGX Orin for many camera streams on one device, larger models, or several models running together.
  • Whichever you pick, train and deploy the model with Roboflow, and move to a bigger module later without retraining.

More on Jetson and Roboflow