ONNX Runtime Plugin for Unity

Pre-built ONNX Runtime libraries for Unity.

See Examples

https://github.com/asus4/onnxruntime-unity-examples

Yolox-Nano

2023-12-25-onnx-yolox.mp4

NanoSAM

2024-02-16-nanosam.mp4

Yolo11 Seg

2025-02-07-yolo11seg.mp4

Supertonic TTS

2026-05-22-supertonic-tts.mp4

Tested environment

• Unity: 6000.6.0f1
• ONNX Runtime: 1.29.0
• ONNX Runtime Extensions: 0.14.0
• ONNX Runtime GenAI: 0.15.2

Execution Providers & Extensions

Execution Providers

Execution Providers are hardware acceleration libraries for each platform. See official docs for more details.

Platform	CPU	CoreML	NNAPI	WebGPU	CUDA	TensorRT	XNNPACK
macOS	✅	✅
iOS	✅	✅					🚧
Android	✅		✅				🚧
Windows	✅			✅	🚧	🚧
Linux	✅				🚧	🚧

Note

Windows GPU uses the WebGPU EP (Direct3D 12) bundled in the core package; DirectML is no longer published by Microsoft. CUDA / TensorRT need the optional win-x64-gpu / linux-x64-gpu package and CUDA 13, cuDNN 9, TensorRT installed. They work in Player builds only.

ONNX Runtime Extensions

ONNX Runtime Extensions are a set of pre/post-processing.

Platform	Extensions
macOS	🚧
iOS	🚧
Android	🚧
Windows	🚧
Linux	🚧

✅ : Supported in Unity Core library
🚧 : Experimental Preview

How to Install

Pre-built libraries are available on NPM. Add the following scopedRegistries and dependencies in Packages/manifest.json.

  "scopedRegistries": [
    {
      "name": "NPM",
      "url": "https://registry.npmjs.com",
      "scopes": [
        "com.github.asus4"
      ]
    }
  ]
  "dependencies": {
    "com.github.asus4.onnxruntime": "0.4.10",
    "com.github.asus4.onnxruntime.unity": "0.4.10",
    "com.github.asus4.onnxruntime-extensions": "0.4.10",
    ... other dependencies
  }

What is included in each package

• com.github.asus4.onnxruntime : Core library
  • CPU provider for all platforms
  • GPU provider for iOS / macOS (CoreML), Android (NNAPI) and Windows (WebGPU)
• com.github.asus4.onnxruntime.unity : (Optional) Utilities for Unity
• com.github.asus4.onnxruntime.win-x64-gpu : (Optional) CUDA / TensorRT provider for Windows x64
• com.github.asus4.onnxruntime.linux-x64-gpu : (Optional) CUDA / TensorRT provider for Linux x64
• com.github.asus4.onnxruntime-extensions : (Optional) ONNX Runtime Extensions

License

Each package is licensed under the MIT License - see the LICENSE file under the package folder for details.