Using Codex CLI with DGrid: A Complete Guide

Introduction

Modern coding agents like Codex CLI enable developers to interact with their codebases using natural language—writing, editing, and reasoning about code directly from the terminal. DGrid AI Gateway provides a unified, OpenAI-compatible API that allows you to access multiple AI models through a single endpoint. By integrating Codex CLI with DGrid, you can route all coding agent requests through a flexible, centralized infrastructure.

This guide walks you through setting up Codex CLI to work seamlessly with DGrid.

Quick Setup

To get started with Codex CLI using DGrid, you’ll need to complete five main steps:

Install and Configure

1. Install Codex CLI following the installation instructions from the GitHub repository
2. Get your DGrid API key from the API Keys page (starts with sk-or-...)
3. Create or edit the config.toml file (typically at ~/.codex/config.toml)
4. Set your API key as an environment variable
5. Start Codex in your project directory

Step 1: Install Codex CLI

Follow the official Codex CLI installation instructions to install the tool on your system. Install globally with your preferred package manager:

Install using npm

npm install -g @openai/codex

Install using Homebrew

brew install --cask codex

Once installed, verify it works

codex --help

If the command runs successfully, you’re ready to proceed. Then simply run codex to get started. Select one of them and then quit Codex.

Step 2: Obtain a DGrid API Key

To use DGrid, you need an API key.

1. Sign in to your DGrid account
2. Generate a new API key
3. Copy and securely store the key(starts with sk-or-...)

DGrid uses API key authentication for all requests through its AI Gateway.

Step 3: Configure Codex CLI to Use DGrid

Codex CLI allows you to define custom model providers via a configuration file.

Create or edit the file: ~/.codex/config.toml

mkdir -p ~/.codex
nano ~/.codex/config.toml

Add the following configuration:

# Default model (fallback if not overridden)

model_provider = "dgrid"
model = "openai/gpt-5.3-codex"
model_reasoning_effort = "high"

[model_providers.dgrid]
name = "DGrid AI Gateway"
base_url = "https://api.dgrid.ai/v1"
env_key = "DGRID_API_KEY"

Step 4: Set Your API Key

You must expose your API key as an environment variable.

macOS / Linux (bash or zsh)

export DGRID_API_KEY="your_api_key_here"

Reload your shell:

source ~/.zshrc
# or
source ~/.bashrc

fish shell

set -Ux DGRID_API_KEY "your_api_key_here"

Windows (PowerShell)

setx DGRID_API_KEY "your_api_key_here"

Verify

echo $DGRID_API_KEY

Core Settings

Step 5: Run Codex

Navigate to your project directory:

cd /path/to/your/project
codex

Codex will now use DGrid as its backend for all model interactions.

Why Use DGrid with Codex CLI

Unified API Access

DGrid provides a single endpoint to access multiple AI providers, simplifying integration and reducing complexity.

OpenAI-Compatible Interface

Codex CLI works seamlessly with DGrid because DGrid implements an OpenAI-compatible API structure.

Flexibility and Control

• Switch models without changing code
• Centralize billing and usage
• Add observability via headers

Production-Ready Infrastructure

DGrid enables you to scale from local experimentation to production deployments with minimal changes.