off-grid-mobile logo

off-grid-mobile

Free

React Native app for running LLMs, vision models, and Stable Diffusion on-device on iOS and Android without internet access. #opensource

FreeFree tier
Inputs: text, image, fileOutputs: text, image
Type
Open Source
Company
Off Grid AI

About off-grid-mobile

Off Grid AI is an open-source React Native app that runs large language models (LLMs), vision models, and Stable Diffusion entirely on-device on iOS, Android, and Mac, with no internet required. It supports text generation (Qwen 3, Llama 3.2, Gemma 3, Phi-4, and custom GGUF files), image generation, vision AI (image analysis), voice transcription, tool calling, and document analysis. All processing happens locally; no data ever leaves the device. An optional Pro tier ($69 lifetime or $49/year) adds on-device text-to-speech (Kokoro), custom personas with persistent memory, and integration with Calendar, email, and MCP servers (Linear, Notion, GitHub) for draft-and-approve actions. Pro also enables secure local network sync between phone and Mac.

Key Features

Runs LLMs, vision models, and Stable Diffusion entirely on-device without internet
Supports text generation with Qwen 3, Llama 3.2, Gemma 3, Phi-4, and any GGUF model
On-device image generation (text-to-image) and vision AI (image analysis)
Voice transcription (speech-to-text) and tool calling capabilities
Document analysis and markdown rendering with streaming responses and thinking mode
Optional Pro: on-device text-to-speech (Kokoro) for hands-free voice interaction
Optional Pro: custom personas with persistent memory and draft-then-approve actions (Calendar, email, MCP servers)
Optional Pro: local network sync between iOS, Android, and Mac

Pros & Cons

Pros
  • Completely offline operation ensures total privacy – no data ever leaves the device
  • Multi-modal: handles text, images, vision, voice, and documents in one app
  • Open-source and free to use with optional affordable Pro tier
  • Supports a wide range of models including custom GGUF files
  • Cross-platform: works on iOS, Android, and Mac with sync (Pro) between devices
Cons
  • Performance limited by device hardware (15–30 tok/s on flagship devices)
  • Voice output (text-to-speech) is only available in the paid Pro version
  • Advanced features like custom personas, tool actions, and network sync require Pro subscription
  • Requires downloading large model files (e.g., multiple GBs) for first-time use
  • No cloud backup or sync of conversations (by design for privacy)

Best For

Personal, privacy-preserving AI assistant for on-the-go queries and tasksGenerating images and analyzing photos without uploading to the cloudSecure document summarization and analysis on-deviceVoice-controlled productivity assistant (with Pro voice mode)Automated drafting of emails, tickets, and replies using connected tools (Pro)Running open-source models locally for experimentation and research

FAQ

Does Off Grid AI require an internet connection?
No, it runs entirely offline. All models and processing happen on-device, and no data ever leaves your device.
What models does Off Grid AI support?
It supports Qwen 3, Llama 3.2, Gemma 3, Phi-4, and any GGUF model. You can also bring your own .gguf files.
Is there a paid version?
The base app is free and open-source. An optional Pro tier ($69 lifetime or $49/year) adds on-device text-to-speech, custom personas, draft-and-approve tool actions, and local network sync.
Can I use Off Grid AI on my Mac?
Yes, there is a dedicated Off Grid AI Desktop app for Mac. A single Pro license covers both your phone (iOS/Android) and your Mac.
What is the performance like?
On flagship devices, you can expect 15–30 tokens per second for text generation, with streaming responses and real-time rendering.