Nativ
A free Mac app that runs AI chatbots like Llama or Mistral directly on your own computer, with no account, no monthly fee, and no data leaving your machine — instead of sending every message to a company's servers.
🔗 Visit Nativ
Description
Using a chatbot like ChatGPT or Claude usually means every message you type gets sent to a company's servers, which raises reasonable questions about privacy and adds up in subscription costs over time. Nativ takes the opposite approach: it runs the AI model directly on your own Mac, so your conversations, code and files never leave your machine.
Nativ is a free, open-source (884+ GitHub stars) native macOS application, built entirely in Swift and SwiftUI, for running open-source AI models locally on Apple Silicon (M1 and newer, macOS 26+ required). It provides a chat interface, model management, real-time system monitoring (per-core CPU load, GPU utilization, unified memory tracking), and both OpenAI-compatible and Anthropic-compatible API server endpoints so other tools (Codex, Claude Code, Pi, Hermes, OpenCode) can connect to your local models. Models are sourced from Hugging Face, including curated partner models from Google, Cohere and Liquid AI. There are no accounts, credits, or data monetization of any kind.
💬 Our review
The short version: Nativ is a clean, genuinely free way to run open-source AI models entirely on your own Mac, and exposing both OpenAI- and Anthropic-compatible API endpoints means it can slot into existing developer tools instead of requiring you to change your whole workflow.
Against cloud AI subscriptions like ChatGPT Plus or Claude Pro, the tradeoff is the standard local-vs-cloud one: Nativ costs nothing ongoing and keeps everything on your machine, but you're limited to whatever open-source models your Mac's hardware can actually run well, which won't match the largest frontier cloud models on raw capability. Against other local-inference tools (Ollama, LM Studio), Nativ's differentiator is being fully native Swift/SwiftUI rather than a cross-platform wrapper, which shows in details like real per-core CPU and GPU monitoring built right into the UI. It's a young project (884 stars, requiring the very recent macOS 26), so expect a smaller community and fewer third-party guides than more established local-inference tools — but for a privacy-conscious Mac user who wants zero subscriptions and full local control, it's a solid, no-cost option.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Entièrement gratuit et open-source, aucun compte ni abonnement requis.
Pros
Entièrement gratuit, aucun compte ni abonnement
100% local : confidentialité totale, aucune donnée envoyée
App native Swift/SwiftUI avec monitoring système intégré
Endpoints API compatibles OpenAI et Anthropic
Cons
Limité aux modèles open-source que le Mac peut exécuter localement
Nécessite macOS 26 récent et Apple Silicon
Communauté plus restreinte que Ollama ou LM Studio