abdulmominsakib/localmind
A mobile application designed to provide a beautiful, fast, and privacy-respecting interface for on device LLM and local LLM servers and cloud providers
What it solves
LocalMind tackles two big problems with typical AI chat apps: privacy and cost. Cloud-based chatbots can track your activity, store your conversations on remote servers, and often require subscriptions. LocalMind keeps everything on your device or talks directly to servers you control, so your data stays private and you pay nothing hidden.
How it works
LocalMind is a mobile app (Android and iOS) built with Flutter. You can either run small language models entirely on your phone — using GGUF (via llama.cpp through llamadart) and LiteRT (via flutter_gemma) — or connect to external backends: Ollama, LM Studio, OpenRouter, Requesty, or any OpenAI-compatible API. The app includes a model manager to download curated models or pull GGUF files from HuggingFace. For chat, it offers a polished interface with streaming responses, markdown and syntax highlighting, expandable reasoning logs for models like DeepSeek R1, editing and regenerating messages, saved prompt templates, message variants, temporary incognito chats, and smart suggested replies. Voice input (speech-to-text) and output (text-to-speech) work natively, including a low-latency voice-to-voice mode. You can attach images and documents for vision-capable models, switch personas on the fly, and swap active models mid-conversation without losing context. All conversation history is stored locally, with full-text search, folders, pinning, and backup/restore via local archives. An optional cloud sync is end-to-end encrypted and only talks to an S3 bucket you own.
Who it’s for
LocalMind suits anyone who wants a private, self-hosted AI chat experience on their phone. It’s especially appealing to privacy-conscious users, developers who already run Ollama or LM Studio on their computers, and tinkerers who want to use custom MCP tools or serve models to other apps via a local API. The polished UI also makes it a good drop-in replacement for mainstream AI chat apps without sacrificing features.
Highlights
- Fully offline on-device inference with GGUF and LiteRT models — no internet needed.
- Connect to Ollama, LM Studio, OpenRouter, Requesty, or any OpenAI-compatible server.
- Built-in Model Context Protocol (MCP) support for custom tools and integrations, with per-chat and global toggles.
- Privacy first: zero analytics, local-only storage, and direct traffic to your own servers.
- Rich chat features: streaming, markdown, reasoning logs, edit/regenerate, message variants, temporary chats, smart replies.
- Voice and audio: speech-to-text dictation, text-to-speech playback with speeds and caching, and real-time voice-to-voice mode.
- Multimodal attachments: send photos, documents, and other files to vision-capable models.
- Manage personas, swap models live, and keep full conversation history with search and folders.
- Backup and restore entire app data locally; optional end-to-end encrypted cloud sync to your own S3 bucket.
- Available in 15 languages, with a dark-first theme and deep customization of chat parameters.
Related
- Project
- Project
- Project
- Project