Codex++ with Nbility
Codex++ is an independent community project for managing providers, protocol conversion, and enhancements around the official Codex desktop app. It is not an official OpenAI client.
Prepare and choose a mode
- Install a matching release from Codex++ Releases and the supported official desktop application.
- Open its manager and verify the app path.
- Create a Nbility provider using your own API credentials.
- Choose Responses or Chat Completions according to model support and your installed release.
The current project README lists both protocols. Older claims that it only supports Chat Completions should not be carried forward.
Enter connection details
| Setting | Value |
|---|---|
| API key | Dedicated key from Token management |
| Standard OpenAI base URL | https://api.nbility.ai/v1 |
| Model ID | An available exact ID |
| Context window | The actual model limit |
Some releases append /v1 in the UI. If the field explicitly requests the service origin, enter https://api.nbility.ai and use diagnostics to verify the final path is /v1/responses or /v1/chat/completions, without /v1/v1.
Test before launching
Run its provider/model test, save, and launch the intended instance from Codex++. Send a short message in a new session and check Usage logs.
Separate instances may use different configuration directories. If an old session uses another service, inspect the active instance’s provider. Verify tools, images, and long contexts separately; protocol conversion does not guarantee every native feature is equivalent.
References
Adapted and rewritten for Nbility from aicentos’s Codex++ guide, licensed under CC BY 4.0. Its account, plan, and proprietary service instructions do not apply here. See the client’s official documentation.