Configure your model API

Using your own model API keeps billing and compliance with your provider. WriteX forwards requests from your machine.

Before you start

Obtain a model API or gateway URL; ensure your network allows the endpoint. Provider walkthroughs: Docs · Get a model API.

Steps

1. Open Settings → Model / Provider.
2. Choose an official provider (e.g. DeepSeek, Xiaomi MiMo, OpenAI) or OpenAI-compatible / custom with your Base URL.
3. Paste the API key and save (encrypted locally).
4. Send a test prompt in the agent.
5. Set assistant output language separately from UI language if needed.

Tips and best practices

Never commit keys to Git. Rotate keys in the provider console when staff changes.

Related reading

• Get a model API
• AI agent feature
• Which models are supported?
• Agent writing workflow