Copy page

Integrations

Codex

Codex is a coding agent developed by OpenAI that you can run locally from your terminal and that can read, modify, and run code on your machine, in the chosen directory. It’s open-source, and built in Rust for speed and efficiency.

Prerequisites

In order to use the AI Gateway with Codex you will need to complete these steps first:

Configure Codex CLI

There are two approaches you can take to using Codex with Zuplo's AI Gateway.

1. Route OpenAI through AI Gateway - This would configure your OpenAI provider in Codex to route requests through the AI Gateway rather than directly to OpenAI

2. Create a Zuplo specific provider - Use Zuplo as a provider that be selected in the Codex configuration and work with any OpenAI compatible models

Configuration steps for both options are below:

Route OpenAI through AI Gateway

Open the Codex configuration ~/.codex/config.toml and modify the model_providers.openai-chat-completions entry so that the base_url points to your app's API URL with /v1 appended, and set your env_key to the API Key of the app you created to use with Codex. The app page in the Zuplo Portal shows the API URL, which ends in the app's ID.

Code
[model_providers.openai-chat-completions]
name = "OpenAI using Chat Completions"
base_url = "https://my-gateway-main-2e18f50.zuplo.app/config_fe0a04972d2848e0a94ae4b8bcd1497e/v1"
env_key = "ZUPLO_AI_GATEWAY_API_KEY"

Save the file and reload Codex. Your OpenAI requests will now be routed through the Zuplo AI Gateway.

Zuplo AI Gateway provider

To add a specific provider for the AI Gateway you can add an additional entry to the Codex config.toml file.

Code
[model_providers.zuplo]
name = "Zuplo AI Gateway"
base_url = "https://my-gateway-main-2e18f50.zuplo.app/config_fe0a04972d2848e0a94ae4b8bcd1497e/v1"
env_key = "ZUPLO_AI_GATEWAY_API_KEY"

Save the file and reload Codex. Your Zuplo AI Gateway provider will now be available and you can switch Codex over to use it at any time by running:

Code
codex --config model_provider="zuplo"

Choosing a model

The app's Model Filtering policy controls which models the app may use. Codex names the model as providerName/model, where providerName is the provider name configured in your gateway:

Code
codex --config model_provider="zuplo" --model "openai/gpt-5"

A request with no model, or a model without that prefix, gets a 400. Adding Model Filtering with an allow list also supplies a default, letting requests omit the model.