Cursor

Configure Cursor chat BYOK with a conditional gateway URL override and documented feature limits.

Install, configure, and test

1. Download the desktop installer for your OS from https://cursor.com/download, install Cursor, and open a project. This guide concerns the editor's provider API keys, not a Cursor CLI account token.

2. In the routerra.ai console (https://routerra.ai/keys), create an API key with access to your intended model and copy its exact model ID. Open Cursor Settings → Models. In the OpenAI API Key field, enter YOUR_API_KEY and click Save (older builds may show Verify). A provider key alone does not redirect requests to the gateway.

3. Only if your installed build exposes Override OpenAI Base URL, enable it and set the base to https://routerra.ai/v1. This is a base URL: do not append /chat/completions. Enable the key if a separate toggle is shown. Add YOUR_MODEL_ID through the custom-model control if available, then select it explicitly in a new chat rather than Auto. Restart Cursor after changing the endpoint.

4. Ask a short question and inspect the gateway request log. Cursor documents BYOK for chat models; its OpenAI integration lists standard, non-reasoning chat models. Tab completion continues using built-in models. Do not assume every Agent feature or arbitrary gateway model works.

If the URL override is absent, this build has no documented gateway setup here. Hidden API-key controls can reflect team policy; ask the administrator to review personal API keys. For rejected models, check the exact ID and the OpenAI key slot; do not rename a model locally unless the gateway provides that alias.