Privacy AI: Powerful chatbot

Privacy AI runs entirely on-device with no backend servers, no accounts, and no data collection. Privacy is built into the architecture. Any network traffic, for remote models only, is fully visible: inspect every request, response, and SSE frame in the built-in Protocol Inspector, or route through mitmproxy for your own verification. In this version, Privacy AI introduces the MLX Audio Engine with on-device TTS, STT, VAD, and speech-to-speech models. Live Transcription enables continuous streaming transcription and real-time translation as audio or video plays, fully on-device with no cloud connection. The AI Gateway now accepts vision inputs, and you can export usage reports from the Cost Estimator. It supports OpenAI, OpenAI Responses, Claude, OpenRouter, and HuggingFace protocols with tool calling. This includes "interleaved thinking" across providers (DeepSeek, Claude, Gemini, Kimi, MiniMax, OpenRouter) to preserve reasoning continuity. The Protocol Inspector logs all protocols, making debugging and auditing fully transparent. A generous "Free Plan" includes offline GGUF/MLX/Apple Foundation models, iCloud sync, Markdown conversion, and "40+ built-in tools". Cloud models, the "MCP Marketplace", and custom API providers require a subscription. - Why Privacy AI Most AI apps lock you into a single service. Privacy AI gives you control: run models fully offline, self-host for compliance, or use cloud APIs when needed. You decide where execution happens, how data flows, and when tools are invoked. - Who It’s For • Developers & infra builders (GGUF, MLX, OpenRouter, custom APIs) • Professionals in law, healthcare, finance, research • Power users building automation workflows • Teams requiring compliance or internal-only AI - What It Does * Run Any Model, Anywhere • Offline on iPhone, iPad, and Mac • Load GGUF/MLX models from HuggingFace or iCloud • Connect to OpenAI, Anthropic, Gemini, DeepSeek, or self-hosted servers • Switch local and remote models mid-chat without losing context • Supports Apple’s on-device Foundation Model (iOS 26) * Advanced Reasoning Control • "Interleaved thinking" for supported reasoning models • "Rethink" button to continue reasoning without re-running tools • "Min-P sampling" for finer control over output focus vs diversity * Use Tools Seamlessly • Full MCP with integrated Marketplace • Call tools (search, stocks, arXiv, HealthKit, JavaScript) offline or remote • Inspect, replay, and debug tool calls directly in chat * Automate Privately • Siri, Shortcuts, and Share Extension integration • Offline TTS and transcription • OpenAI protocol TTS (gpt-4o-tts, gpt-4o-mini-tts) * Files On-Device • Convert PDFs, Office files, EPUB, HTML, YouTube, audio • OCR and clean Markdown conversion • Read "vCard (.vcf)" and "iCalendar (.ics)" files • Export to Markdown, PDF, HTML, EPUB, or JSON * Images & Editing • Generate images locally or via OpenRouter (e.g. Gemini 2.5 Flash) • Supports gpt-image-1 and FLUX.1-dev • Annotate with Apple Pencil or touch * Content & Reading • Retrieve and convert "X.com article URLs" into clean Markdown • Optimized streaming Markdown rendering for long responses * Control & Extend • “Natural Talk” voice UI • Live token and context usage indicator • Custom prompts, tools, and preferences • iCloud sync for models, chats, and settings * Built for Apple • Optimized for M-series Macs and iPad Pro • Requires iOS 18.6+ • Deep Siri, Shortcuts, and Share Extension integration - No Account. No Surprises. No forced sign-ups. Choose local, self-hosted, or cloud. Your models. Your device. Your data. Terms of Use: https://privacyai.acmeup.com/docs/policy/tos.html Privacy Policy: https://privacyai.acmeup.com/docs/policy/privacy.html