Codex Deployment Guide
Covers CLI installation, API access, and development environment configuration.
📋 Prerequisite (Step One)
Complete the Node.js installation first and make sure Node.js 18+ is installed correctly. When configuring CC-Switch, remember to open the advanced options. You can start chatting after adding the model.
🚀 Quick Setup with CC-Switch (Recommended)
⚠️ Prerequisite: Before configuring Codex with CC-Switch, make sure you have installed the Codex tool globally with npm:
npm install -g @openai/codex@latest
Verify the installation: codex --version
We recommend the CC-Switch quick configuration tool for simple graphical setup without command-line work.
Configuration Steps
- Launch CC-Switch and switch to the Codex tab
- Open the CC-Switch application
- Click the "Codex" tab at the top
- Click the orange "+" button in the upper-right corner to add a configuration
- Enter the Codex provider configuration
- Provider name: Enter a custom name, such as "gpt-agent"
- Base URL: Enter
http://137.184.200.132:3001/api/v1 - API Key: Paste your API key
- Model: Select gpt-5.4 (the latest model when this guide was written was 5.4)
- Other settings: Adjust reasoning effort, network access, and other parameters as needed
- Click "Save"
💡 Tip:
- CC-Switch automatically creates the
~/.codex/config.tomland auth.json files - You can add multiple provider configurations and switch between them at any time
- After switching configurations, close and restart Codex for the change to take effect
Enable and use the configuration:
- Find the configuration you just created in the configuration list
- Click "In Use" on the right side of the configuration
- The configuration will be marked "In Use" with a green label
- Restart Codex to apply the new configuration
Switch quickly from the system tray: CC-Switch lets you switch Codex configurations from the system tray:
- Right-click the CC-Switch icon in the system tray
- Select the Codex category from the menu
- Select the configuration you want to use
- The configuration takes effect immediately without opening the main window
⚠️ Important:
- You must restart Codex after switching configurations
- You can test API endpoint speed in CC-Switch
⌨️ Manual Command-Line Setup
If you do not use CC-Switch, follow the steps below to configure Codex manually.
🖥️ Windows
Step One: Install the Codex Command-Line Tool
Launch Command Prompt or PowerShell as an administrator, then run:
npm install -g @openai/codex@latest codex --version
Create the configuration file: config.toml
model_provider = "gpt-agent" model = "gpt-5.4" model_reasoning_effort = "xhigh" disable_response_storage = true approval_policy = "on-request" sandbox_mode = "danger-full-access" model_supports_reasoning_summaries = true [model_providers.gpt-agent] name = "ikun" base_url = "http://137.184.200.132:3001/api/v1" wire_api = "responses" requires_openai_auth = true
Create the authentication file: auth.json
{
"OPENAI_API_KEY": "paste-your-codex-api-key-here"
}Step Three: Initialize the Workspace
mkdir my-codex-project cd my-codex-project codex
🍏 macOS
Install the Codex Tool
npm install -g @openai/codex@latest codex --version
Integrate the API Service
Create the configuration directory:
mkdir -p ~/.codex cd ~/.codex
Create the config.toml configuration:
cat > config.toml << 'EOF' model_provider = "gpt-agent" model = "gpt-5.4" model_reasoning_effort = "xhigh" disable_response_storage = true approval_policy = "on-request" sandbox_mode = "danger-full-access" model_supports_reasoning_summaries = true [model_providers.gpt-agent] name = "ikun" base_url = "http://137.184.200.132:3001/api/v1" wire_api = "responses" requires_openai_auth = true EOF
Create the auth.json authentication configuration:
cat > auth.json << 'EOF'
{
"OPENAI_API_KEY": "paste-your-codex-api-key-here"
}
EOFInitialize the Workspace
mkdir my-codex-project cd my-codex-project codex
🐧 Linux
Install the Codex Tool
sudo npm install -g @openai/codex@latest codex --version
Integrate the API Service
Create the configuration directory:
mkdir -p ~/.codex cd ~/.codex
Create the config.toml configuration:
cat > config.toml << 'EOF' model_provider = "gpt-agent" model = "gpt-5.4" model_reasoning_effort = "xhigh" disable_response_storage = true approval_policy = "on-request" sandbox_mode = "danger-full-access" model_supports_reasoning_summaries = true [model_providers.gpt-agent] name = "ikun" base_url = "http://137.184.200.132:3001/api/v1" wire_api = "responses" requires_openai_auth = true EOF
Create the auth.json authentication configuration:
cat > auth.json << 'EOF'
{
"OPENAI_API_KEY": "paste-your-codex-api-key-here"
}
EOFInitialize the Workspace
mkdir my-codex-project cd my-codex-project codex
Where Are the Configuration Files Stored?
- Windows:
%USERPROFILE%\.codex\ - macOS/Linux:
~/.codex/
Configured but Unable to Chat?
Try the steps below. If the issue persists, contact support.
- Verify your Base URL is exactly
http://137.184.200.132:3001/api/v1 - Confirm your API key is valid and has not expired
- Check that your proxy settings in CC-Switch are correctly configured
- Try testing the API endpoint speed in CC-Switch to confirm connectivity