CoreAI-Model-Zoo
A catalog of downloadable, verified model conversions and their conversion recipes for Apple's on-device Core AI runtime on iPhone and Mac, each carrying the recipe and validation evidence used to produce it.
CoreAI-Model-Zoo converts PyTorch models into Apple Core AI's .aimodel bundle format for iOS 27 and macOS 27, the successor to Core ML, so they run on the GPU or Neural Engine. Every model is gated against its original source model before it ships, and its recipe is published so it can be rerun or the converted bundle re-verified against the source.
Each card records what was measured, on which hardware, and how strong that model's parity proof is, since verification differs per model; where the shipped configuration could not be recovered, the recipe says so instead of guessing. Community members can contribute their own model ports from their own Hugging Face namespace, credited by name. A companion Swift package, CoreAIKit, downloads and caches a chosen model and runs prompts locally, with a bundled chat demo app and CLI.