PreferredBackend enum

Hardware backend for model inference.

Platform support:

• cpu: All platforms
• gpu: All platforms (Metal on macOS, DirectX on Windows, Vulkan on Linux, OpenCL on Android)
• npu: With LiteRT-LM (.litertlm models) — Android (Qualcomm Snapdragon only: the Android archive ships the Qualcomm QNN dispatch stack and nothing for MediaTek or Google Tensor) and Windows (Intel LunarLake/PantherLake). Asking for it elsewhere falls back rather than throwing; read InferenceModel.activeBackend for what actually ran. On Windows the check is per OS, not per device: a PC without an Intel NPU is still offered npu, and has been measured reporting activeBackend == npu while running elsewhere. There, npu is a request, not a proof.

gpu and npu fall back if unavailable — GPU, then CPU. cpu does not fall back: it is the last resort already.

Inheritance

• Object
• Enum
• PreferredBackend

Available extensions

• EnumName

Values

cpu → const PreferredBackend

gpu → const PreferredBackend

npu → const PreferredBackend

Properties

hashCode → int

The hash code for this object.

no setter, inherited

index → int

A numeric identifier for the enumerated value.

no setter, inherited

name → String

Available on Enum, provided by the EnumName extension

The name of the enum value.

no setter

runtimeType → Type

A representation of the runtime type of the object.

no setter, inherited

Methods

noSuchMethod(Invocation invocation) → dynamic

Invoked when a nonexistent method or property is accessed.

inherited

toString() → String

A string representation of this object.

inherited

Operators

operator ==(Object other) → bool

The equality operator.

inherited

Constants

values → const List<PreferredBackend>

A constant List of the values in this enum, in order of their declaration.