everyapi
Integration · Codex desktop managerOpenAI-compatible · one key · live model catalog
Codex

Run the EveryAPI gateway inside Codex++

Codex++ launches and manages the Codex desktop app. Create a pure API provider for EveryAPI, select the Responses API protocol, then start Codex through Codex++.

01Setup code

One base URL line points Codex++ at EveryAPI

Default global Base URL / China-first Base URL: https://api.everyapi.ai/v1 / https://api-cn.everyapi.ai/v1

Codex++ Manager · Providers
# Codex++ Manager → Provider configuration → Add provider
# Mode: Pure API
Base URL: "https://api.everyapi.ai/v1"
Protocol: "Responses API"
Model: "gpt-5.6-luna"
API Key: "sk-everyapi-..."
 
# Save, switch to this provider, then launch Codex++
02Steps

From zero to your first call

  1. Create a Pure API provider

    Open Codex++ Manager, add a provider, and select Pure API. This keeps the EveryAPI key separate from any official Codex login.

    Codex++ Manager → Providers → New provider

  2. Paste the EveryAPI connection

    Enter EveryAPI's /v1 base URL, your EveryAPI key, and a model from your account catalogue. Select Responses API as the protocol, then save and switch to this provider.

  3. Start Codex through Codex++

    Launch the desktop app from the Codex++ entry point, not the original app shortcut. Codex++ loads the saved provider before starting Codex.

03FAQ

Codex++ × EveryAPI

Why does the provider test say "Stream must be set to true"?

Codex++'s provider test sends a non-streaming request. EveryAPI's Codex route requires stream: true, so that test can return HTTP 400. Save and switch the provider, then verify with a streamed request or by launching Codex through Codex++.

Why select Responses API instead of Chat Completions?

The Codex desktop app uses the Responses API. Selecting Responses API keeps its native request format intact; Codex++ handles provider configuration and startup.

Can I choose another model?

Use a model listed in your EveryAPI account catalogue. The example is only a starting point; available models and routes depend on your account.