Client Connection

Change one setting at a time so the result identifies the cause.

1. Establish a baseline

Run the first request. If curl fails, fix the key, group, model, or network first. If curl succeeds, focus on the client.

2. Check the resulting URL

A normal OpenAI-compatible combination is:

Base URL: https://sprelaytoken.com/v1
Client appends: /chat/completions
Final URL: https://sprelaytoken.com/v1/chat/completions

The three lines show base, appended path, and final result. Common malformed results are:

https://sprelaytoken.com/v1/v1/chat/completions
https://sprelaytoken.com/v1/chat/completions/chat/completions

3. Check active configuration

Confirm that the current session uses the new provider, restart the client or terminal, check whether environment variables override UI settings, and search for duplicate provider names in multiple config files.

4. Check model and network

The actual Model ID must exactly match the model API and be available in the key's group. Confirm that the browser can reach https://sprelaytoken.com, the system clock is correct for TLS, and local or company proxies do not rewrite or block the connection.

No usage log usually means the request never reached Sprelay. A failed usage log means the service received it; continue with API Errors.