API keys
Markdown Locker does not sell model tokens. You bring a key from the provider you want; the app sends requests from your machine.
That choice is deliberate. AI models improve fast, and prices move. Locking you into a single vendor’s meter would mean paying twice — once for the app, again for inference you could buy cheaper or run locally. Instead, Think and Tidy use your key. You pick Grok, OpenAI, OpenRouter, a local Ollama model, or any OpenAI-compatible endpoint. When a better model ships, you switch without waiting on us.
Your notes never become our training data. The key stays on your Mac. Only the context needed for a given question is sent to the provider you chose, the same way any other client would.
Where to paste a key
Open Think, then the gear icon. Paste the key, pick a provider and model, and confirm. A green status dot means a key is saved. You can change or clear it anytime.
Supported providers
| Provider | Base URL | Notes |
|---|---|---|
| Grok (xAI) | https://api.x.ai/v1 |
Default. Models such as grok-3 |
| OpenAI | https://api.openai.com/v1 |
Standard OpenAI API keys and models |
| OpenRouter | https://openrouter.ai/api/v1 |
One key, many models (e.g. x-ai/grok-3) |
| Ollama | http://localhost:11434/v1 |
Local models — no cloud key |
| Custom | You set it | Any OpenAI-compatible chat endpoint |
Getting a key
Create an account with your chosen provider, generate an API key in their console, and paste it into Markdown Locker. Usage and billing stay with that provider — we never see your invoice.