@orqo/foundationmodels-runtime

Apple Foundation Models provider adapter — daemon-backed implementation of IntelligenceProvider.

Install

pnpm add @orqo/foundationmodels-runtime

Usage

This package provides the AppleFoundationModelsProvider and CoreAIModelsProvider adapters that wire the IntelligenceProvider port from @orqo/foundationmodels-core to the local Swift daemon via @orqo/foundationmodels-client.

It is registered automatically when you call createFoundationModels() in @orqo/foundationmodels. Direct instantiation is useful when you need to configure the daemon adapter in isolation or register it alongside other providers.

import { createAppleFoundationModelsProvider } from "@orqo/foundationmodels-runtime";
import { createFoundationModels } from "@orqo/foundationmodels";

// Instantiate the daemon-backed provider directly.
const appleProvider = createAppleFoundationModelsProvider();

// Inject into the FoundationModels runtime alongside any custom providers.
const fm = await createFoundationModels({
  providers: {
    custom: [appleProvider],
  },
});

const result = await fm.respond({ input: "Hello from the runtime layer." });
console.log(result.content);

Also exports createCoreAIModelsProvider for the CoreAI backend variant.

Most applications should depend on @orqo/foundationmodels rather than this package directly.

License

Apache-2.0 — see LICENSE and NOTICE. Maintained by Orqo.