Omni – Local-first multimodal file search on macOS

Omni is a local-first macOS app that uses a state-of-the-art omni embedding model to index and search across text, PDFs, images, audio, and video entirely on-device, built with a Swift-native UI and mlx-swift-transformer core (no Python). It is designed for privacy-conscious power users and developers who need fast, multimodal file retrieval without cloud dependency, tested on Apple Silicon from M3 Pro to M4 Pro. The project is interesting because it delivers a fully local, cross-modal search experience that was previously impractical, leveraging Apple’s hardware efficiently while avoiding the complexity of Python-based ML stacks.