RunAnywhere Binaries

Pre-built XCFrameworks for the RunAnywhere on-device ML inference framework.

Installation

Swift Package Manager (Recommended)

Add to your Package.swift:

dependencies: [
    .package(url: "https://github.com/RunAnywhereAI/runanywhere-binaries.git", from: "1.0.0")
]

Then add the backend you need to your target:

.target(
    name: "MyApp",
    dependencies: [
        // Choose one or more backends:
        .product(name: "RunAnywhereONNX", package: "runanywhere-binaries"),
        // .product(name: "RunAnywhereCoreML", package: "runanywhere-binaries"),
        // .product(name: "RunAnywhereTFLite", package: "runanywhere-binaries"),
    ]
)

Or in Xcode: File → Add Package Dependencies → Enter repository URL.

CocoaPods

Add to your Podfile:

# Default ONNX backend
pod 'RunAnywhere'

# Or specific backend(s)
pod 'RunAnywhere/ONNX'
pod 'RunAnywhere/CoreML'
pod 'RunAnywhere/TFLite'

# All backends
pod 'RunAnywhere/All'

Then run:

pod install

Manual Download

Download XCFrameworks from Releases:

1. Download the .xcframework.zip for your chosen backend
2. Verify the checksum: shasum -a 256 -c checksums.txt
3. Unzip and drag into your Xcode project
4. In Build Phases → Link Binary With Libraries, add:
   • Foundation.framework
   • CoreML.framework (for ONNX/CoreML)
   • Accelerate.framework
   • Metal.framework (for CoreML/TFLite)

Available Backends

Backend	Use Case	Size*
ONNX	General purpose, cross-platform models	~50MB
CoreML	Apple Neural Engine optimization	~5MB
TFLite	TensorFlow models, Android parity	~20MB

*Sizes are approximate and vary by version.

Choosing a Backend

• ONNX Runtime (Recommended): Best compatibility, supports most model formats
• CoreML: Best performance on Apple devices, requires CoreML model format
• TFLite: Use when you need Android/iOS parity with TensorFlow models

Supported Platforms

• iOS 15.0+
• macOS 12.0+

Version History

See CHANGELOG.md for release notes.

License

MIT License - see LICENSE for details.

Links

• RunAnywhere SDK Documentation
• API Reference
• Examples