← All integrations

RavenRouter × Codex

Codex CLI Integration

Codex CLI uses the OpenAI Responses API via POST /v1/responses on RavenRouter.

Before you start

  • Create an API key in Console → API Keys.
  • Pick a model from GET /v1/models that your plan allows.

API base URL: https://api.ravenrouter.ai/v1

Setup steps

  1. Admins must configure a Codex upstream channel (OAuth).
  2. Create an API token.
  3. Edit ~/.codex/config.toml and point model_provider base_url to https://api.ravenrouter.ai/v1.
  4. Set api_key via env var or config.
  5. Use gpt-5.x-codex models enabled on your Codex channel.

Connection summary

# OpenAI Responses API (Codex CLI)
Base URL: https://api.ravenrouter.ai/v1
API Key:    <YOUR_RAVENROUTER_TOKEN>
Model:      gpt-5.2-codex
Endpoint:   POST /v1/responses

Configuration example

# ~/.codex/config.toml
model = "gpt-5.2-codex"

[model_providers.ravenrouter]
name = "RavenRouter"
base_url = "https://api.ravenrouter.ai/v1"
env_key = "RAVENROUTER_API_KEY"

# shell
export RAVENROUTER_API_KEY="<YOUR_RAVENROUTER_TOKEN>"

Notes

  • Codex only supports /v1/responses and /v1/responses/compact, not chat/completions.
  • The gateway passes Codex CLI headers and sets OpenAI-Beta when needed.
  • Codex models require an admin-configured Codex channel.