Witsy

A closer look at Witsy

Witsy is a desktop AI assistant and universal MCP client from Kochava-Studios on GitHub: it is a BYOK, bring-your-own-keys, application, so you need API keys for the LLM providers you want to use, or you can use Ollama to run models locally on your machine for free. It claims to be the first of very few universal MCP clients, letting you run MCP servers with virtually any LLM, and it supports chat, image creation, video creation, and text-to-speech across providers including OpenAI, Anthropic, Google, xAI, Meta, Mistral, DeepSeek, OpenRouter, Groq, and Cerebras, under an AGPL-3.0 license.

Editorial guidance based on published documentation. The workflow below is a suggested starting point, not a measured product result.

When to choose Witsy

• Power users who want one desktop assistant across many models, since Witsy supports many providers for chat, image, video, and speech, plus MCP servers with virtually any LLM.
• Privacy-minded users who want local models, since Witsy works with Ollama to run models locally on your machine for free.

When to choose something else

• Users who want models included in a single subscription, since Witsy is BYOK and requires your own API keys for hosted providers, with model costs paid to those providers.
• Users who want a web-only tool with no desktop install, since Witsy is a desktop application, installable from its releases page or, on macOS, with brew install --cask witsy.

A practical starting workflow

1. Download Witsy from the releases page, or on macOS install it with brew install --cask witsy.
2. Add API keys for the LLM providers you want to use, or set up Ollama to run models locally for free.
3. Chat with your chosen engine and model, and use image creation, video creation, and text-to-speech across providers.
4. Run MCP servers with virtually any LLM, and use the CLI, API, and agent webhooks for automation once the desktop app is running.

Check before committing

• Witsy itself carries no app price in the repository text read, and it is AGPL-3.0 licensed; hosted model use is paid to providers through your own keys, so no app price or free allowance fact is recorded in this pass.
• The universal MCP client claim, that Witsy is the first of very few universal MCP clients, is a repository claim with a question mark in the original text, not an independent comparison.
• The repository text read includes extensive API and CLI documentation, which is developer documentation, not a beginner tutorial.

Common questions

Compare approaches for this task

• Whisperstream — People who want dictation without sending voice to the cloud, since Whisperstream transcribes entirely on a Windows PC or Apple Silicon Mac and says your audio is never sent to the cloud for core transcription.; Fast thinkers who type slower than they speak, since Whisperstream's pitch is that typing caps you at 40 to 60 words a minute while you think and speak roughly three times faster.
• Wispr Flow — People who write in many apps and hate typing, since Wispr Flow turns speech into clear, polished writing in every app, on Mac, Windows, iPhone, and Android.; Teams that need notetaking plus admin control, since paid plans add unlimited Notetaker access, team management, admin controls, and support.