ReyLocalAI

Private AI Assistant — 100% On-Device

Your conversations never leave your device. ReyLocalAI runs powerful AI models directly on your iPhone or iPad with zero cloud dependency. Upload documents, ask questions, and get intelligent answers — all completely offline and private.

Your AI. Your device. Your privacy.

ReyLocalAI is a fully private AI assistant that runs entirely on your iPhone or iPad. Powered by state-of-the-art Qwen3 language models and Apple's MLX framework, every conversation and every document stays on your device — nothing is ever sent to a server.

Key Features:

100% On-Device AI

All AI processing happens locally on your device. No cloud APIs, no subscriptions, no data collection. After downloading a model once, the app works completely offline.

Document Intelligence

Upload PDFs, text files, markdown, CSV, and more. ReyLocalAI reads, indexes, and summarizes your documents so you can ask questions and get instant answers grounded in your files.

Personalized Memory

The AI remembers your name, preferences, and facts you share — all stored locally. Get responses tailored to you without sacrificing privacy.

Multiple AI Models

Choose from three model sizes to match your device:

  • Qwen3 0.6B — Ultra-fast, works on all devices
  • Qwen3 1.7B — Balanced quality and speed
  • Qwen3 4B — Highest quality for capable devices
Conversation History

Your sessions are saved on-device. Pick up where you left off anytime.

Fine-Tuned Controls

Adjust temperature, sampling, and response length. Toggle the AI's internal reasoning visibility for transparency into how it thinks.

Zero Data Collection

No analytics. No tracking. No telemetry. No ads. We literally cannot see your data because it never leaves your phone.

Built with MLX Swift by 402 Labs Inc.

Categories

  • Primary: Productivity
  • Secondary: Utilities

Keywords

AI, private, offline, on-device, assistant, chatbot, LLM, documents, PDF, privacy, local, no-cloud, MLX, Qwen

What's New in Version 1.0

  • Fully on-device AI assistant powered by Qwen3 models
  • Document upload and intelligent Q&A (PDF, TXT, MD, CSV, JSON, XML, HTML)
  • Pre-processing and summarization for fast document queries
  • User memory and personalization
  • Three model sizes to choose from
  • Adjustable generation parameters
  • Retro CRT terminal interface
  • Complete offline functionality after initial setup