Use apimodels in OpenAI Codex

Get running in 3 steps

1. 1

   Grab an API key

   Open Console in your console and create an sk_… key. New accounts get $1 of free credit.

2. 2

   Drop the config

   Paste the snippet below into ~/.codex/config.toml (create the file if needed).

3. 3

   Export + run

   Export your key, then just run codex.

Full config

# ~/.codex/config.toml
model_provider = "apimodels"
model          = "gpt-5.4"

[model_providers.apimodels]
name     = "apimodels"
base_url = "https://apimodels.app/api/v1"
wire_api = "responses"
env_key  = "APIMODELS_API_KEY"

export APIMODELS_API_KEY="sk_…your_key…"
codex

Settings reference

Reasoning effort

Bump reasoning depth by appending a suffix to the model name. All levels share the same per-token price; higher effort just emits more reasoning_tokens (billed as part of output_tokens).

30-second sanity check

Before installing Codex, confirm the endpoint and your key work with a single curl:

curl -s https://apimodels.app/api/v1/responses \
  -H "Authorization: Bearer $APIMODELS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4",
    "input": "Reply with exactly: ok",
    "max_output_tokens": 16
  }'

Expected response (truncated):

{
  "id": "resp_...",
  "object": "response",
  "model": "gpt-5.4",
  "status": "completed",
  "output": [{
    "type": "message",
    "role": "assistant",
    "content": [{ "type": "output_text", "text": "ok" }]
  }],
  "usage": {
    "input_tokens": 22,
    "output_tokens": 5,
    "total_tokens": 27
  }
}

HTTP 200 with output[0].content[0].text === "ok" means you're good to go.

Why not wire_api = "chat"?

Codex supports both wire_api modes, but GPT-5.4 specifically is only routed via /v1/responses upstream — /chat/completions can't reach it. Other gpt-* models (gpt-5, gpt-5.1, gpt-5.2) work with either; for those you can use wire_api = "chat" against /v1/chat/completions. For GPT-5.4 it must be responses.

Troubleshooting