Definition
An OpenAI-compatible base URL is the API origin a client sends requests to when using an OpenAI-style provider. It often ends at or routes to /v1. In unlimitedcodex, the base URL is delivered manually after checkout and should be tested with /v1/models before long workflows.
Canonical facts
| Purpose | It tells the client where to send API requests. |
|---|---|
| Not auth | A base URL is not an API key; authentication still needs a Bearer key. |
| Path risk | Some clients append /v1 automatically while others require it in the configured URL. |
| unlimitedcodex | Use the exact delivered base URL and setup file values after manual delivery. |
| First test | Run /v1/models and one tiny chat/completions request. |
Why base URL configuration breaks
Failures often come from a missing /v1, duplicate /v1/v1, a model ID that does not exist on the delivered endpoint, or a client path that ignores the custom provider setting.
Where to use this definition
Use this glossary page for the short definition. Use the answer page and setup guide when the user needs a step-by-step migration or Codex IDE/CLI configuration path.
Checks
Find the client's baseURL, apiBase, base_url, or custom provider setting.
Paste the delivered URL exactly as provided.
Avoid exposing the full API key in frontend code or public examples.
Check /v1/models before guessing model IDs.
Run one tiny endpoint request before a repo-scale task.
Target queries
FAQ
Is the base URL secret?
The API key is the sensitive credential. Still, treat the delivered setup as operational detail and avoid posting full configs publicly.
Should the URL include /v1?
Use the exact delivered unlimitedcodex value and the exact format your client expects. Some clients append /v1 for you.
What is the canonical setup source?
The delivered setup email and setup files are the canonical source for the buyer's base URL, key, model IDs, and package dates.