off-grid-ai/off-grid-ai-mobile

The Swiss Army Knife of Offline AI. Chat, Speak, and Generate Images - Privacy First, Zero Internet. Download an LLM and use it on your mobile device. No data ever leaves your phone. Supports text-to-text, vision, text-to-image

What it solves

Off Grid AI provides a complete, private AI suite that runs entirely on-device, eliminating the need for cloud connectivity and preventing data from leaving the user's phone or Mac. It solves the problem of privacy-conscious users needing access to multimodal AI capabilities (text, image, vision, and voice) without relying on external servers.

How it works

The application leverages native hardware acceleration (such as NPUs on Snapdragon and Core ML on iOS) and optimized libraries like llama.cpp and whisper.cpp to run models locally. It supports GGUF models for text generation and Stable Diffusion for images. For knowledge base functionality, it uses a bundled MiniLM model for on-device embedding and SQLite for local storage and retrieval via cosine similarity.

Who it’s for

Privacy-focused individuals, developers, and users on Android, iOS, and macOS who want a versatile, offline AI assistant that can handle multiple modalities of AI tasks.

Highlights

  • Multimodal Capabilities: Supports text generation (Llama, Qwen, Phi), image generation (Stable Diffusion), vision AI (SmolVLM, Qwen-VL), and voice transcription (Whisper).
  • On-Device RAG: Includes a project knowledge base where PDFs and text documents are embedded and retrieved locally.
  • Tool Calling: Built-in tools for web search, calculators, and knowledge base search with runaway prevention.
  • Local Server Connectivity: Ability to connect to OpenAI-compatible local servers like Ollama or LM Studio.
  • Hardware Optimized: NPU-accelerated image generation and real-time voice transcription.