Use with Codex
[!WARNING] We recommend CC Switch for managing and switching Codex provider configurations — no manual environment-variable or config-file editing needed.
[!INFO] This page is for the official Codex CLI (command line). If you use the Codex desktop app with the third-party enhancement tool Codex++, see Use with Codex++ instead.
Codex CLI supports adding custom model providers through its config file. This is a good fit for saving yylx.io as a standalone provider you can then select directly inside Codex.
[!INFO] The config file suits long-term use. For a quick test, you can just set the environment variable and specify the model with a command-line flag.
What you need
| Field | Value |
|---|---|
| Provider ID | yylx |
| Provider Name | yylx.io |
| Base URL | https://app.yylx.io/v1 |
| API Key environment variable | YYLX_API_KEY |
| Model | A model name from the console's model list |
First, open the "API Keys" page in the yylx.io console and create or pick a Key dedicated to Codex. Including codex in the name makes usage easier to trace later.
Set the API Key
Set the Key in your terminal first:
export YYLX_API_KEY="sk-your-api-key"
For long-term use, add it to ~/.zshrc or ~/.bashrc.
If you added it to a shell config file, remember to run:
source ~/.zshrc
Confirm the current terminal can read the variable:
echo $YYLX_API_KEY
Any non-empty output is fine — don't paste the full Key into public chats or issues.
Configure the Codex provider
Edit ~/.codex/config.toml and add:
model = "your-model-name"
model_provider = "yylx"
[model_providers.yylx]
name = "yylx.io"
base_url = "https://app.yylx.io/v1"
env_key = "YYLX_API_KEY"
wire_api = "responses"
Replace your-model-name with a model name shown in the yylx.io console.
Field reference
| Field | Description |
|---|---|
model_provider |
The provider ID used by default |
base_url |
The yylx.io OpenAI-compatible address |
env_key |
The environment variable Codex reads the API Key from |
wire_api |
The API style Codex uses; follow what your Codex version supports |
[!TIP] If your Codex version's config fields differ slightly from the example, keep the same core pieces: provider ID, Base URL, environment variable name, and model name. After upgrading Codex, it's also worth re-checking the config file format.
Launch
codex
To switch models temporarily, use Codex's launch flags or its in-session model picker. On first launch, start with a lightweight request — for example, ask the model to print its configured model name or a one-line summary — to confirm the provider is active.
Verify requests reach yylx.io
Check both sides:
| Where | What to look for |
|---|---|
| Codex terminal | Normal responses; any authentication or unknown-model errors |
| yylx.io usage log | Whether the Codex request you just made appears |
If nothing shows up in the yylx.io usage log, Codex usually isn't using the yylx provider, or the current terminal can't read YYLX_API_KEY.
Troubleshooting
For authentication errors, first confirm the terminal can read the variable:
echo $YYLX_API_KEY
For unknown-model errors, make sure the model field matches the console name exactly.
For network or API-format errors, check that base_url stops at /v1 — not /v1/chat/completions or /v1/responses.
[!TIP] If you switch between providers often, let CC Switch manage the yylx.io configuration to cut down on manual edits to
~/.codex/config.toml.