Nativ

Nativ is a free, open-source native macOS app for running AI models entirely locally on Apple Silicon, with no cloud, accounts, or subscriptions involved. It works as a private chat app, a local model manager, a live performance dashboard, and an OpenAI/Anthropic-compatible local inference server that other tools can point to as if it were a cloud API. Built around Apple's MLX framework, Nativ bundles an mlx-vlm server, automatically finds compatible models already in a user's Hugging Face cache (respecting HF_HUB_CACHE/HF_HOME), and wraps the experience in a native SwiftUI interface rather than a browser-based dashboard. It runs open language, vision, video, code, and embedding models, and includes live monitoring of per-core CPU load, GPU utilization, unified memory and swap pressure, disk throughput, storage capacity, SMART drive health, and thermal/power sensors, so users can see exactly what running a local model costs the machine in real time. It installs via a DMG from GitHub Releases or through Homebrew.