IntegrationsDev tool
Codex CLI
Run OpenAI Codex CLI against RouteMux by overriding the base URL and key.
Codex CLI speaks the OpenAI protocol, so RouteMux is a drop-in base URL swap.
Configure
export OPENAI_BASE_URL="https://api.routemux.com/v1"
export OPENAI_API_KEY="sk-..." # your RouteMux keyThis base URL does keep /v1
Unlike the Anthropic variable, the OpenAI base URL includes /v1 — OpenAI clients
append only the operation path (/chat/completions). Getting these two backwards is the
single most common misconfiguration.
Pick a model
export OPENAI_MODEL="openai/gpt-5.5"GET /v1/models returns exactly the models your key can call — treat it as the source of
truth rather than a published catalogue.
Verify
curl https://api.routemux.com/v1/chat/completions \
-H "Authorization: Bearer $OPENAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"openai/gpt-5.5","messages":[{"role":"user","content":"ping"}]}'Notes
- Streaming works unchanged; RouteMux forwards SSE without buffering.
- Billing is per real usage, settled after the upstream finishes. Timeouts are not billed.
- A
402means an empty wallet, not a bad request.