Kuzco

Kuzco is a local AI runtime for iOS that lets you run large language models, vision models, and image generation directly on-device. Its primary purpose is to give mobile developers a simple, privacy-preserving way to integrate advanced AI features into their apps without relying on remote APIs or recurring usage fees. With a minimal Swift API, Kuzco is designed to be easy to adopt and efficient to run on modern Apple hardware.

Kuzco supports running LLMs for tasks like chat, summarization, code assistance, and content generation, as well as computer vision models for image understanding, classification, and object detection. It also enables on-device image generation, allowing apps to create or transform images without sending user data to external servers. The SDK is optimized for iOS, handling model loading, execution, and memory management so developers can focus on product logic rather than ML infrastructure. By running everything locally, Kuzco reduces latency, improves responsiveness, and removes dependency on network connectivity or third-party API limits.