On-device AI Models for Swift on iOS and macOS
Drop-in Swift packages that run state-of-the-art models on device, on iOS and macOS, in a few lines of code on Core ML.
Every available Desert Ant Labs model ships as an idiomatic Swift package: async/await, Core ML, and a clean API that feels like an extension of the platform's own SDK. Small models are bundled in the app; larger ones download once from Hugging Face and cache locally.
Add the package in Xcode and call it. Inference stays on device, so audio, images, and text never leave the device.
Available models
Reversible PII redaction across 24 EU languages, from a ~13 MB Core ML model bundled in the app.
Emoji suggestions from text in under 10 ms, small enough to run on every keystroke.
Content topics from any text across 101 languages, from a ~74 MB Core ML model, or a ~15 MB English-only build.
PencilKit snap-to-shape in one call, from a 0.2 MB Core ML model, no download.
DeepFilterNet 3 speech enhancement on Core ML: a 5-minute clip in about 2.6 seconds, 114x realtime.
Frame-precise filler-word detection straight from the waveform, no transcript.
Language identification for short text across 84 languages, from a 2 MB model bundled in the app.
Available soon
Pricing
Every model is free up to 100k monthly active devices per SDK. Unlimited inference per user.