> ## Documentation Index
> Fetch the complete documentation index at: https://docs.toapis.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Codex Setup Guide

> Use ToAPIs in Codex CLI through CC Switch

## Prerequisites

Before you begin, prepare:

1. **A ToAPIs API key** from the [ToAPIs Console](https://toapis.com/console/token).

2. **Codex CLI**, installed from the [official Codex documentation](https://developers.openai.com/codex/cli) or with:

   ```bash theme={null}
   npm install -g @openai/codex
   ```

3. **CC Switch**, downloaded only from [ccswitch.io](https://ccswitch.io) or the [official GitHub Releases](https://github.com/farion1231/cc-switch/releases).

## Add ToAPIs in CC Switch

1. Open CC Switch and select **Codex** in the application switcher.
2. Click **+** and choose **App-specific provider**.
3. Select the **Custom** preset and enter:

| Field                       | Value                   |
| --------------------------- | ----------------------- |
| Name                        | `ToAPIs`                |
| API Key                     | Your ToAPIs API key     |
| API endpoint / Base URL     | `https://toapis.com/v1` |
| Model                       | `gpt-5.3-codex`         |
| API protocol                | `Responses`             |
| Local routing/model mapping | **Off**                 |

4. Click **Add**, then click **Enable** on the ToAPIs provider card.
5. Exit any running Codex session and reopen the terminal so the new configuration is loaded.

<Note>
  ToAPIs natively supports `POST /v1/responses`. Do not enable local routing/model mapping; that option is intended for providers that only expose Chat Completions.
</Note>

## Start and Verify

```bash theme={null}
codex
```

Send a simple task, for example:

```text theme={null}
Inspect this project and summarize its main directory structure.
```

If Codex responds normally, the setup is complete.

## Change the Model

Edit the ToAPIs provider to use another Responses-compatible model available to your account. CC Switch can fetch models from `/v1/models`; you can also review the [ToAPIs model list](../api-reference/chat/models).

Common choices:

| Model                    | Best for                                                        |
| ------------------------ | --------------------------------------------------------------- |
| `gpt-5.3-codex`          | Recommended default for most coding tasks                       |
| `gpt-5.3-codex-spark`    | Coding tasks that prioritize response speed                     |
| `gpt-5.3-codex-official` | Workloads that specifically require the official direct channel |

Availability depends on your account and channel configuration. Prefer the `/v1/models` list fetched by CC Switch.

## Troubleshooting

### Codex still uses the previous provider

Codex does not hot-reload provider configuration. Fully exit Codex, reopen the terminal, and run `codex` again.

### `401 Unauthorized`

* Confirm the API key in CC Switch is complete
* Verify that the key is active in the [ToAPIs Console](https://toapis.com/console/token)
* Ensure the account has sufficient balance

### `404` or an incorrect Responses path

Confirm the Base URL is `https://toapis.com/v1`, the protocol is `Responses`, and local routing/model mapping is disabled.

### Model unavailable

Use CC Switch's model-fetch button or call `GET https://toapis.com/v1/models` with the same API key.

## Configuration and Key Safety

CC Switch manages `~/.codex/auth.json` and `~/.codex/config.toml`. Do not commit these files, your API key, or screenshots containing the key to Git.

See the [official CC Switch user manual](https://github.com/farion1231/cc-switch/tree/main/docs/user-manual/en) for additional provider-management instructions.