Install Netron on Ubuntu or Linux after checking your package architecture, then use the Debian or RPM desktop release, Python package, or browser app.
Choose the right Netron installation method
Netron displays an Open Neural Network Exchange (ONNX), TensorFlow Lite, Keras, or PyTorch model as a graph, while the current Netron release page publishes a Debian package for amd64 systems and an RPM package for x86_64 systems, so check your architecture before downloading either file.
dpkg --print-architecture
| Your system | Best method | Why |
|---|---|---|
| Ubuntu or Debian on amd64 | Debian package | Adds the desktop application through APT |
| Fedora, RHEL, or a compatible x86_64 distribution | RPM package | Installs the desktop application through DNF |
| arm64 Linux or a remote server | Python virtual environment | Runs Netron in a local web server without an architecture-specific desktop package |
| A machine where you cannot install software | Browser app | Opens a supported model from the Netron web interface |
If you want the browser route, open Netron in your browser and select a local model. Your browser still has to load and parse the file, so large models can take longer than small samples.
Install Netron on Ubuntu or Debian
Use this path only when dpkg reports amd64. The release currently has no arm64 Debian asset, so forcing its amd64 package onto an arm64 system fails at the package architecture boundary.
Download the Debian file from the official Netron release, open a terminal in the download directory, and install it with APT.
sudo apt install ./Netron-*.deb
APT reads the local Debian package and resolves its shared-library dependencies from your configured repositories. If you want more context on that process, see the Linux package management guide.
Launch Netron from your desktop menu or run the following command. The application should open at its model picker.
netron
Select Open Model, choose a supported file, and wait for the graph to render. A model with external weight files must keep those files beside the main model so Netron can resolve them.
Install the RPM package on Fedora or RHEL
The release page also provides an x86_64 RPM. Download it, open a terminal in the same directory, and let DNF install the package plus available dependencies.
sudo dnf install ./Netron-*.x86_64.rpm
Run netron from the terminal or open it from your application menu. The desktop workflow is otherwise the same as the Debian package.
Install Netron with Python
The Netron Python package works when the desktop release does not match your architecture, and it keeps the installation inside a disposable virtual environment. I installed the current package in a fresh environment and served a SqueezeNet ONNX model over localhost.
python3 -m venv netron-env
netron-env/bin/python -m pip install --upgrade pip netron
Confirm which version the environment installed.
netron-env/bin/netron --version
Download a small ONNX model or use one you already have. The Netron model list provides sample files for several supported formats.
netron-env/bin/netron squeezenet1.0-12.onnx \
--host 127.0.0.1 --port 8080
Open http://127.0.0.1:8080 in the same machine’s browser. Binding to 127.0.0.1 keeps the viewer local. If you need remote access, use an SSH tunnel instead of exposing the port directly.
The graph pane shows operators and tensor flow. Select a node to inspect shapes, attributes, and model metadata in the side panel.
Fix common Netron installation problems
Most failures come from a package format or architecture mismatch rather than Netron itself. Check these conditions before changing repositories or forcing an install.
- APT reports an architecture error. Run dpkg –print-architecture. Use the Debian package only on amd64, or switch to the Python method.
- The netron command is missing after the Python install. Run it from netron-env/bin/netron because a virtual environment does not add itself to every shell session.
- A model opens without its weights. Keep external tensor files in the location expected by the main model file.
- You find Snap instructions. The Netron Snap listing is deprecated and points Linux users to Debian or RPM packages. The Snap-to-Deb package comparison explains the practical difference between those formats.
- The download command is unfamiliar. Use the cURL download guide or the wget command guide before automating release downloads.
For another machine-learning desktop tool that also separates desktop and headless paths, compare the LM Studio installation guide. The same architecture check prevents wasted package downloads.
Frequently asked questions
These answers cover the choices that change the installation path.
Can I install Netron on Ubuntu with APT?
Yes. Download the current amd64 Debian package from the official Netron release page, then install the local file with sudo apt install ./Netron-*.deb.
Does Netron support arm64 Linux?
The Python package runs on arm64 with a supported Python interpreter. The current desktop release provides amd64 Debian and x86_64 RPM packages, so arm64 users should choose Python or the browser app.
Should I install Netron with Snap?
No. The Netron Snap listing is deprecated and directs Linux users to Debian or RPM packages.
How do I open an ONNX model in Netron?
Launch the desktop app and select Open Model, or run netron-env/bin/netron followed by the model filename to serve it on localhost.
Verify the installation
Open a model and select one node. A working installation shows the graph, tensor shapes, operator attributes, and model properties. Keep the Python environment if you use Netron regularly, or delete the netron-env directory when you no longer need it.