Cursor

Cursor

Cursor can send chat and agent requests to LivRouter when you override its OpenAI base URL with your own endpoint and key. Cursor's settings UI changes between versions, so treat the labels below as a map, not an exact transcript.

Setup

  1. Create a LivRouter key on the API keys page.
  2. In Cursor, open Settings → Models (the API key section; older versions label it API Keys).
  3. Enable the Override OpenAI Base URL toggle and set it to https://api.livrouter.com/v1.
  4. Paste your LivRouter key into the OpenAI API key field and apply.
  5. Add the models you plan to use by their exact IDs from the model catalog — for example glm-5.3 or claude-sonnet-4-6 — and enable them.

Verify the setup

Open a chat in Cursor, pick one of the models you added, and send a short prompt. A normal reply plus a matching entry in your LivRouter usage logs confirms the requests are routing through your account.

What does not run through LivRouter

Features that execute on Cursor's own infrastructure — Tab autocomplete, Apply, and other Cursor-hosted model features — are not served by custom endpoints. Only the models you explicitly add and select go through LivRouter. Plan usage in Cursor around chat and agent conversations with those models.

Troubleshooting

  • Verification fails or 401 — re-copy the key from the API keys page; it must be a LivRouter key, not an OpenAI account key.
  • Model not found — model IDs must match the catalog exactly. Re-copy the ID from the catalog; IDs change when the catalog updates.
  • Requests ignore the override — make sure the selected model in the chat is one you added under the custom endpoint, not a Cursor-default model.

For the raw request format behind the chat, see the API quickstart.