# Core AI Framework Lab

A SwiftUI workbench for discovering, converting, running and benchmarking machine-learning models with Apple's CoreAI framework, covering language, vision, audio and diffusion model recipes.

Core AI Framework Lab is a native macOS and iOS application for exploring Apple's CoreAI.framework, the asset, specialization and runtime layer that sits alongside but is distinct from Core ML or Apple's FoundationModels API; only one bundled model adapter bridges through FoundationModels. It exposes workspaces for browsing pinned Apple model recipes and their export commands, converting a model on macOS with preflight checks and live logs, authoring and validating recipe and pipeline contracts without executing them, inspecting a model's metadata and compute types, running task-specific playgrounds such as a bundled text-to-speech pipeline and an experimental speaker-diarization pipeline, and authoring iPhone deployment profiles.

It records provenance, benchmark evidence, physical-device evidence and Swift integration exports for each project so results can be reproduced and checked.

It is written in Swift, built with Xcode 27, and released under the MIT license.

Website: https://developer.apple.com/core-ai/