InferenceMode

@PublicPreviewAPI
public final class InferenceMode

---

Specifies how the SDK should choose between on-device and in-cloud inference.

Summary

Nested types
public static class InferenceMode.Companion

Public fields
static final @NonNull InferenceMode	ONLY_IN_CLOUD Only use in-cloud inference.
static final @NonNull InferenceMode	ONLY_ON_DEVICE Only use on-device inference.
static final @NonNull InferenceMode	PREFER_IN_CLOUD Prefer in-cloud inference, but fallback to on-device if cloud is unavailable.
static final @NonNull InferenceMode	PREFER_ON_DEVICE Prefer on-device inference, but fallback to in-cloud if unavailable.

Public fields

ONLY_IN_CLOUD

public static final @NonNull InferenceMode ONLY_IN_CLOUD

Only use in-cloud inference.

ONLY_ON_DEVICE

public static final @NonNull InferenceMode ONLY_ON_DEVICE

Only use on-device inference.

PREFER_IN_CLOUD

public static final @NonNull InferenceMode PREFER_IN_CLOUD

Prefer in-cloud inference, but fallback to on-device if cloud is unavailable.

In this mode, the SDK will use in-cloud inference only unless the device is offline, at which point the SDK will fall back to on-device inference.

PREFER_ON_DEVICE

public static final @NonNull InferenceMode PREFER_ON_DEVICE

Prefer on-device inference, but fallback to in-cloud if unavailable.

In this mode, the SDK will attempt to use on-device inference first and, if the on-device model is unable to generate an answer, it will retry using the cloud model.