muxd AI Agent vs DeviceAI

muxd is a mobile client for the muxd AI agent an open-source daemon that runs on your machines and works with any model from any provider. Connect to your agent by scanning a QR code or entering a host address. Manage multiple machines through a single server, switch between nodes, and resume any session from your phone. WHAT YOU CAN DO • Chat with your AI agent from anywhere • Send images for analysis and visual tasks • Browse and resume sessions across all your machines • Monitor node health, uptime, and connection status • Switch between machines instantly through the central server • Pick up conversations exactly where you left off MULTI-MACHINE HUB Run one serverto coordinate all your muxd daemons desktop, laptop, server, VM. The app connects to the server and lets you pick which machine to talk to. Node health updates in real time with heartbeat monitoring. ANY MODEL, ANY PROVIDER muxd works with Claude, GPT, Mistral, Grok, Fireworks, Ollama, and any OpenAI-compatible API. Your choice of model, your API key, running on your hardware. BUILT-IN TOOLS Your agent has 27 tools including file operations, bash commands, web search, git, scheduling, project memory, and more. Describe what you need and the agent picks the right tool. PRIVATE BY DESIGN Everything runs on your machines. No cloud middleman. Sessions are stored locally in SQLite. API calls go directly to your chosen provider. Zero telemetry. Zero data collection. OPEN SOURCE muxd is free and open source under the Apache 2.0 license. The full codebase is on GitHub. REQUIRES A running muxd daemon or hub on your network. Install muxd on your machine from https://muxd.sh.

Run powerful AI models entirely on your iPhone — no internet, no subscriptions, no data sharing. DeviceAI is a privacy-first chat app that runs large language models (LLMs) completely on your device using Apple's Metal GPU acceleration. Your conversations never leave your phone. KEY FEATURES • 100% On-Device Processing — All AI inference happens locally. Zero cloud dependency. • 60+ Open-Source Models — Choose from Llama, Qwen, Phi, Mistral, DeepSeek and more • Full Privacy — No data collection, no analytics, no telemetry. Your chats stay on your device. • Offline-First — Works perfectly without internet after model download • Streaming Responses — See AI responses appear in real-time • Reasoning Models — Support for advanced thinking models like DeepSeek R1 • Markdown & Code — Beautiful formatting with syntax highlighting • Customizable — Adjust system prompts, temperature, and generation parameters • Conversation History — All chats saved locally with Isar database PERFECT FOR • Privacy advocates who want AI without cloud exposure • Travelers and remote workers who need offline AI assistance • Developers testing open-source models • Students wanting unlimited, free AI tutoring • Anyone who values data sovereignty TECHNICAL DETAILS • Powered by llamadart (llama.cpp for Flutter) • Supports GGUF format models from HuggingFace • Metal GPU acceleration for fast inference (15-40 tokens/second) • Models range from 0.5B to 8B parameters • Requires iOS 16.0 or later Download models once, use forever. No subscriptions, no API keys, no compromises. Your data stays on your device. Always.