Set up Codex API access with CodingRouter
Manage the Codex model provider, API key and model from CodingRouter Connect. Review changes to the local configuration before switching, with backups available for recovery.
API configuration and account sign-in are different
Install Codex separately before using Connect. Signing in to your CodingRouter account lets Connect read the tokens and models available to you. The token you select supplies the API key used by Codex; your account login is not itself a model API key. Existing Codex login information should not be substituted for a missing provider key.
Select the token and model in Quick setup
Choose Codex, open Quick setup and select a usable token. Check its group, then select a model from the allowed list. Existing valid choices can be retained. If you create a token, confirm its quota and expiry first. A model that does not appear in the allowed list should not be assumed to work with your token.
Understand what the confirmation changes
Connect prepares Codex's CodingRouter model_provider, base_url and model settings in its TOML configuration, together with the selected API key in its authentication configuration. The provider uses the Responses route. You do not need to paste a config.toml snippet or hand-edit auth.json. Review the current provider and expand the actual file paths before applying, especially if you have an existing setup.
Resolve a conflict or restore a previous setup
If the preview expires or a file changes, generate a new preview before confirming. Do not bypass a missing-key error with existing login credentials. Restore a previous setup from the saved backups if needed, then reopen Codex so it reads the configuration. “Configured” confirms a local write; it does not claim a successful or free model invocation.
Set up your AI coding tool
Practical guides for API access, model selection and configuration recovery.
