Nativ is a 100% open-source, MIT-licensed macOS-native AI app designed for Apple Silicon. Powered by the MLX framework, it runs frontier open models locally, supporting chat, code completion, voice transcription, and more. Completely free — no account registration, no subscription, no cloud connection needed.

Nativ Free Local AI Desktop App: Run Open Models on Apple Silicon - Zero Subscription, Zero Cloud, Zero Accounts

What is Nativ

Nativ is a macOS native app built with SwiftUI, using MLX-VLM (Meta's MLX inference framework) as the inference engine. It's not just a chat interface — it's a complete local AI workspace.

Core features:

Local chat & vision — Streaming responses, image attachments, reasoning output
Model library management — Discover installed MLX models, one-click download, switch models
Performance analytics — Track requests, token usage, TTFT, decode speed
Local API server — OpenAI-compatible /v1/chat/completions endpoint
Coding agent integration — Codex, Claude Code, Pi, Hermes, OpenCode
Menu bar controls — Start/stop server, switch models, view stats

System Requirements

Item	Requirement
Hardware	Apple Silicon Mac (M1 or later)
OS	macOS 26+ (Sequoia)
Unified Memory	Depends on model size (see below)
License	MIT (completely free)
Network	Only needed for initial model download, fully offline at runtime

Supported Models

Nativ comes pre-integrated with curated open models from Google, Cohere, and Liquid AI:

Model	Source	Context Window	Model Size	Capabilities
Gemma 4 E2B Instruct	Google	128K	10.28 GB	Vision + Audio
North Mini Code	Cohere	500K	19.38 GB	Code + Tool Use
LFM2.5-VL 1.6B	Liquid AI	128K	3.20 GB	Vision + Language

Memory requirements:

3.2 GB model: Requires at least 8GB unified memory
10.28 GB model: Requires at least 16GB unified memory
19.38 GB model: Requires at least 24GB unified memory (M2/M3 Max or M4 Pro/Max)

Why Nativ Matters

1. True Local Processing — Zero Data Leakage

All inference runs on your Mac. Your prompts, conversation history, and image attachments never leave your device.

2. Completely Free, MIT License

Free download and use with no usage limits. Anyone can review the source code, fork and modify. No "enterprise edition" traps.

3. Local API Server

Starts a local API server at http://127.0.0.1:8080 by default, supporting both OpenAI and Anthropic compatible formats.

4. Coding Agent Integration

Built-in support for Codex, Claude Code, Pi, Hermes, and OpenCode.

Installation

Method 1: Direct Download (Recommended)

Visit Nativ GitHub Releases page
Download the latest Nativ-*.dmg file (~342 MB)
Drag Nativ to Applications folder
On first launch, select an installed model or continue on-demand

Method 2: Build from Source

brew install xcodegen
make xcode-generate
make xcode-build
open build/XcodeDerivedData/Build/Products/Debug/Nativ.app

Comparison with Cloud AI Services

Feature	Nativ (Local)	ChatGPT	Claude	Gemini
Cost	✅ Completely free	$20/mo	Pay-per-use	Partial free
Privacy	✅ Data stays on device	❌ Data uploaded	❌ Data uploaded	❌ Data uploaded
Offline	✅ Fully offline	❌ Requires internet	❌ Requires internet	❌ Requires internet
Custom models	✅ Switch any MLX model	❌ Fixed model	❌ Fixed model	❌ Fixed model
API compatibility	✅ OpenAI + Anthropic	—	✅ Anthropic	✅ OpenAI compatible
Coding agent integration	✅ Built-in	❌	❌	❌

Use Cases

Privacy-sensitive conversations — Medical, legal, finance professionals can use local models to ensure sensitive information never leaves the device
Coding assistance — Configure coding agents to use local models for Copilot-like experience without paying
Offline work — Use without internet connection
Model research — Performance analytics panel for monitoring different models

FAQ

Is Nativ really completely free?

Yes, MIT license. Free download and use with no usage limits.

Does it require internet?

Only for initial model download. Fully offline at runtime.

Does it support Intel Mac?

No, Apple Silicon only (M1/M2/M3/M4).

Can I switch between different models?

Yes, download more MLX format models from Hugging Face.

Summary

Nativ is the most complete local AI solution for Apple Silicon. Completely free and open-source, no accounts or subscriptions needed, supports multiple models and coding agent integration. The best choice for users who value privacy and don't want to rely on cloud services.

Project: github.com/tylerneely/nativ

Data current as of July 21, 2026. Free policies may change.

Nativ Free Local AI for Mac: Run Open Models on Apple Silicon — Zero Subscription, Zero Cloud, Zero Accounts