Connect AI1API to Codex CLI

Codex CLI uses the same .codex configuration path and Responses protocol as Codex Desktop.

Prerequisites

Codex Desktop is installed
An API Key exists and is bound to an available group

Setup steps

Step 1: Create an API Key

Open API Keys in the AI1API console and create a key for the group you want to use. Never share the full key in tickets, screenshots, or repositories.

Step 2: Configure Codex

Use ~/.codex/config.toml on macOS/Linux or %USERPROFILE%\.codex\config.toml on Windows. The console Use API Key dialog can generate the file.

~/.codex/config.toml

model_provider = "ai1api"
model = "gpt-5.6-sol"
review_model = "gpt-5.6-sol"
model_reasoning_effort = "xhigh"

[model_providers.ai1api]
name = "AI1API"
base_url = "https://ai1api.com/v1"
wire_api = "responses"
requires_openai_auth = true

Step 3: Add the API Key

Replace the placeholder with your own key and keep it in a local config file only.

~/.codex/auth.json

{
  "OPENAI_API_KEY": "sk-your-key"
}

Verify the connection

Start with the smallest request, then enable streaming and tools after the connection works.

Minimal Responses check

curl https://ai1api.com/v1/responses \
  -H "Authorization: Bearer $AI1API_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-5.6-sol","input":"Reply with OK","stream":false}'

Notes

Use the model and group shown by live pricing and the console.
A 403 usually means that the key is not bound to a group that allows Responses.