OpenAI

OpenAI

OpenAI is a first-party model provider — it produces the GPT family and serves them from its own API. It is the zero-config default for the codex llm tool, and a first-class model provider: once connected, operator lists its available models live for delegators to pick from.

Model provider ≠ llm tool. OpenAI (the provider) serves the models; Codex (the llm tool) is the CLI. A delegator pairs a tool with a provider’s model.

Connect

Operator references your key by env-var name — it never stores the secret:

export OPENAI_API_KEY="sk-..."

A provider is connected when its /models probe succeeds. OpenAI then shows ● connected in the Model Providers view with its live model list.

Listing models

Operator probes https://api.openai.com/v1/models and lists whatever the API returns (agnostic to the specific model set):

GET /api/v1/model-servers/kinds/openai-api/models   # { reachable, models[], error? }

Use from a delegator

[[delegators]]
name = "codex-gpt"
llm_tool = "codex"
model = "gpt-4o"
# model_server omitted → resolves to the implicit openai-api default

Custom endpoint / proxy

Declare an openai-api server with an explicit base_url to point Codex at a proxy; that base URL is then injected at spawn (OPENAI_BASE_URL). The probe default (https://api.openai.com) is probe-only — it never changes the launch path on its own.