Overview

Witsy is an open-source, bring-your-own-keys desktop AI assistant that works as a universal Model Context Protocol (MCP) client. It supports 15+ AI providers including OpenAI, Anthropic, Google Gemini, xAI, Mistral, DeepSeek, Ollama, LM Studio, and OpenRouter. Features include multi-modal chat with vision, text-to-image and text-to-video generation, document RAG, real-time voice chat, system-wide AI commands for selected text, and agent webhooks for automation.

Pricing: Free and open source

Architecture: Apple Silicon, Intel

Key Features

15+ AI providers — OpenAI, Anthropic, Gemini, xAI, Mistral, DeepSeek, Ollama, LM Studio, OpenRouter, and more
Universal MCP client — run MCP servers with any LLM
Text-to-image and text-to-video generation
Image editing and image-to-video capabilities
Document chat (RAG) with local files
Real-time voice chat mode
Speech-to-text dictation and text-to-speech
System-wide AI commands for highlighted text in any app
Prompt Anywhere for content generation in any application
Long-term memory plugin
Anthropic Computer Use support
HTTP API and agent webhooks for automation
Bring your own API keys — no subscription fees