Codex
Configure Codex for your enterprise and verify the request in your console logs.
Configuration
Use your enterprise HTTPS gateway URL, an API key scoped to this tool and a model ID that supports the protocol. The address below comes from this gateway’s public configuration. Find an eligible model
- Merge this custom provider into ~/.codex/config.toml. Replace the gateway URL and the ID of a model that supports Responses.
- Use your enterprise secret manager to inject TOKEN_PYXIS_API_KEY into the environment of the restarted Codex process.
| API protocol | Address to enter |
|---|---|
| Responses | Base URLhttps://tokenpyxis.com/v1 |
model = "YOUR_RESPONSES_MODEL_ID"
model_provider = "token_pyxis"
[model_providers.token_pyxis]
name = "Token Pyxis"
base_url = "https://tokenpyxis.com/v1"
env_key = "TOKEN_PYXIS_API_KEY"
wire_api = "responses"
supports_websockets = falseVerification
Send a text request in a new session, then test Responses streaming and tool calls in a test project.
Check the model, timestamp and token usage in the enterprise console logs. If the tool shows a request ID, look it up directly. See reconciliation steps
Troubleshooting
If authentication fails, check the launching process environment. End the base URL at /v1 so the client appends /responses. This configuration uses HTTP.
