LocalAI: Offline AI Chat Agent

Delta Tech Sol

Contains adsIn-app purchases

About this app

LocalAI is a 100% offline AI chatbot that runs open-source LLM models directly on your phone. No internet. No account. No data collection. Your chats, prompts and photos never leave your device.

Turn your Android into a private AI assistant that works on a plane, on the metro, in the mountains — anywhere with zero signal.

🔒 PRIVATE AI, 100% ON-DEVICE
• Offline AI chat — every token is generated locally by an optimized llama.cpp engine
• No sign-up, no API key, no subscription needed to start
• Zero telemetry, zero tracking, zero cloud processing
• Airplane-mode proof: an AI assistant offline that never phones home

🧠 BUILT-IN AI MODEL HUB (GGUF)
Download, switch and manage open-weight LLM models offline — sized for real phones, from 3GB RAM budget devices to flagships:
• Meta Llama 3.2 1B & Llama 3.2 3B Instruct
• Google Gemma 2 2B Instruct
• Alibaba Qwen2.5 0.5B & Qwen2.5 1.5B Instruct
• Microsoft Phi-3 Mini & Phi-3.5 Mini Instruct
• DeepSeek R1 Distill Qwen 1.5B
• SmolLM2 360M & SmolLM2 1.7B Instruct
• TinyLlama 1.1B Chat
Every model card shows file size, RAM requirement and what it is good at — so you never download an AI model your phone cannot run.

🖼️ OFFLINE VISION AI — CHAT WITH IMAGES
Attach a photo, screenshot or diagram and ask questions about it, processed 100% offline:
• SmolVLM 500M, SmolVLM2 256M & SmolVLM2 500M for fast image description
• Gemma 3 4B Vision for high-quality image understanding on flagship devices
Describe photos, read text in images, explain screenshots and charts — no internet required.

💻 OFFLINE CODING ASSISTANT
• Qwen2.5 Coder 0.5B, 1.5B, 3B and 7B
• Write, explain, refactor and debug code offline
• Syntax-highlighted code blocks with one-tap copy

🧩 OFFLINE REASONING AI
DeepSeek R1 Distill and Phi-3.5 Mini think step by step through math, logic and problem solving — entirely on-device.

💬 A REAL CHAT EXPERIENCE
• Local chat history stored privately on your device
• Markdown rendering + syntax-highlighted code
• Adjustable temperature, context length and system prompt
• Background model downloads that resume
• Clean Material Design UI with light and dark themes

⚡ WHAT PEOPLE USE LOCALAI FOR
• Offline AI writer — emails, essays, captions, rewrites and grammar fixes
• Offline AI summarizer — turn long text into clear key points
• Offline AI translator for travel with no roaming or WiFi
• Offline study helper — explanations, quizzes and revision notes
• Private journaling, brainstorming, roleplay and story writing
• Coding help on a train, a plane or a remote site with no signal

📱 DEVICE REQUIREMENTS
• 3–4GB RAM: light models (SmolLM2 360M, Qwen2.5 0.5B, SmolVLM2 256M)
• 6–8GB RAM: balanced models (Gemma 2 2B, Llama 3.2 3B, Phi-3.5 Mini)
• 10GB+ RAM: Qwen2.5 Coder 7B and Gemma 3 4B Vision
• Storage: roughly 300MB to 5GB per model
Speed is hardware-dependent — Snapdragon 8 Gen 2+ or Dimensity 9000+ devices generate noticeably faster tokens per second. The first run of a model is slower while it loads into memory.

Download LocalAI today and run a private AI assistant offline — free, unlimited and completely yours.

LocalAI works as an offline AI app, a local AI chatbot and an on-device AI assistant in one: run LLM models offline, chat with AI without internet, and use offline AI tools for writing, summarizing, translation, study, coding and image analysis. Local LLM manager, GGUF model downloader, offline vision AI, offline coding AI and offline reasoning AI — no WiFi, no login, no ads on your data. AI offline, private AI, secure on-device AI processing.

What’s new

- Fixed minor bugs and crashes.