Cursor Custom OpenAI Base URL Setup with Sprelay
Cursor settings vary by version and account plan. The goal is to direct OpenAI-compatible requests to Sprelay.
- Create a dedicated
Cursorkey with a small quota. - Open Cursor Settings and find Models or API Keys.
- Enter the Sprelay key in the OpenAI API Key field.
- Enable Override OpenAI Base URL, or the equivalent setting.
- Enter
https://sprelaytoken.com/v1. - Add an exact model ID from the live catalog.
- Save and test with a short chat or code explanation.
| Field | Value |
|---|---|
| OpenAI API Key | Your dedicated Sprelay key |
| OpenAI Base URL | https://sprelaytoken.com/v1 |
| Model ID | Exact ID supported by the key's group |
If the actual request path contains /v1/v1, the current Cursor version is appending /v1; use the site root only after confirming that behavior in the error or network log. Chat, Agent, and completion features may use separate model settings, so test them individually.
Model selection and feature limits
The Sprelay catalog includes gpt-6.1-sol and gpt-6-astra on OpenAI-compatible endpoints. A listing in Sprelay does not guarantee that the installed Cursor version accepts that model for every feature. Select a compatible group and test a short chat first.
Cursor's custom API key support has feature restrictions; review the official API key documentation for your version and plan. Do not assume that configuring a key enables Tab completion or every Agent feature.
Confirm requests reach Sprelay
After a test, inspect Sprelay usage logs for the expected model, key, status and charge. If no request appears, check Cursor's selected provider and the actual request URL. If a request appears but fails, follow the recorded status in API errors.
Use a minimal API request to distinguish a key or group problem from a client configuration problem. Keep a small key quota until you understand the billing dimensions.
