---
id: netron
version: "9.2.2"
license: MIT
license_treatment: permissive
maintenance: active
---

# netron — Viewer for neural network, deep learning and machine learning models.

License: permissive · Maintenance: active · Downloads: 98.6K/mo

## What it is and what it does

Netron is a standalone viewer for inspecting and visualizing neural network and machine learning model architectures. It supports a wide range of model formats spanning multiple frameworks—ONNX, TensorFlow, PyTorch, Keras, Core ML, OpenVINO, and others—making it useful for developers and researchers who need to understand model structure, layer composition, and data flow without running inference.

The package can be used as a command-line tool (run `netron [FILE]` after installation), accessed through a browser at netron.app, or called programmatically via Python. It has no runtime dependencies, making installation and use lightweight. The project is actively maintained and widely used, as evidenced by its popularity and recent releases.

Use it for:

- Inspect ONNX model architecture and layer details during model development or debugging
- Visualize PyTorch or TensorFlow model graphs to understand data flow and layer connections
- Examine pre-trained models from model zoos to understand their structure before fine-tuning
- Compare model architectures across different frameworks or versions
- Explore Keras or TensorFlow Lite models for mobile or edge deployment planning
- Validate Core ML or OpenVINO models before deployment on target platforms

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Netron is a viewer for neural network and machine learning models that supports ONNX, TensorFlow Lite, PyTorch, TensorFlow, Core ML, OpenVINO, Keras, Caffe, Darknet, Safetensors, NumPy, and several other formats.

Yes. Netron is a lightweight, actively maintained, dependency-free viewer for a broad range of ML model formats. It solves a genuine need for developers and researchers who need to inspect model structure without writing code. The MIT license is permissive, there are no known vulnerabilities, and the tool is widely adopted. Install it if you work with neural networks or machine learning models in any of the supported formats.

## Install

pip install netron
uv add netron
poetry add netron

## Installing netron

Before you install: Installation is straightforward with no runtime dependencies. The package is actively maintained with recent releases and a large repository following (33351 stars), indicating stable, well-supported tooling.

License in practice: MIT license permits unrestricted use, modification, and distribution with minimal restrictions, making this suitable for both personal and commercial projects.

Quickstart:

pip install netron
netron path/to/model.onnx

Or programmatically:
import netron
netron.start('path/to/model.onnx')

Verify before relying:

- Whether the Python API (netron.start) is fully documented and stable across versions
- Performance characteristics when opening very large model files
- Whether experimental format support (MLIR, JAX, GGUF, RKNN, ncnn, MNN, PaddlePaddle, scikit-learn) is production-ready

## Package facts

- License: MIT (permissive)
- Python support: unspecified
- Install friction: low
- Maintenance: active
- Downloads: 98.6K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

neural network model viewer, deep learning model visualization, onnx model inspector, pytorch model visualizer, tensorflow model viewer, machine learning model explorer, model architecture inspector, model-visualization, neural-networks, ml-debugging

[View on SkillFed](https://skillfed.io/packages/netron) · [View on PyPI](https://pypi.org/project/netron/)