What causes "OpenAI API key is not configured" in invoke-ai/InvokeAI?

Set external_openai_api_key (or its environment variable) to a valid OpenAI API key and restart InvokeAI.

invoke-ai/InvokeAI · error · ExternalProviderRequestError

OpenAI API key is not configured

Error message

OpenAI API key is not configured

What it means

ExternalProviderRequestError raised by OpenAIProvider.generate when the app config has no external_openai_api_key set. The provider uses the key for the Authorization: Bearer header on /v1/images/generations and /v1/images/edits, and skips making a request that would certainly fail with 401. This is a local configuration error, not an API response.

Source

Thrown at invokeai/app/services/external_generation/providers/openai.py:34

    ExternalGenerationResult,
)
from invokeai.app.services.external_generation.image_utils import decode_image_base64


class OpenAIProvider(ExternalProvider):
    provider_id = "openai"

    _GPT_IMAGE_MODELS = {"gpt-image-1", "gpt-image-1.5", "gpt-image-1-mini", "gpt-image-2"}
    _DEFAULT_TIMEOUT = 120
    _MODEL_TIMEOUTS: dict[str, int] = {"gpt-image-2": 300}

    def is_configured(self) -> bool:
        return bool(self._app_config.external_openai_api_key)

    def generate(self, request: ExternalGenerationRequest) -> ExternalGenerationResult:
        api_key = self._app_config.external_openai_api_key
        if not api_key:
            raise ExternalProviderRequestError("OpenAI API key is not configured")

        model_id = request.model.provider_model_id
        is_gpt_image = model_id in self._GPT_IMAGE_MODELS
        timeout = self._MODEL_TIMEOUTS.get(model_id, self._DEFAULT_TIMEOUT)
        size = f"{request.width}x{request.height}"
        base_url = (self._app_config.external_openai_base_url or "https://api.openai.com").rstrip("/")
        headers = {"Authorization": f"Bearer {api_key}"}

        use_edits_endpoint = request.mode != "txt2img" or bool(request.reference_images)

        opts = request.provider_options or {}

        if not use_edits_endpoint:
            payload: dict[str, object] = {
                "model": model_id,
                "prompt": request.prompt,
                "n": request.num_images,
                "size": size,

View on GitHub (pinned to 0b6a024f2f)

Solutions

Set external_openai_api_key (or its environment variable) to a valid OpenAI API key and restart InvokeAI.
Guard before dispatch: call provider.is_configured() and show a configuration error to the user instead of attempting generation.
Confirm the key is visible inside the runtime environment (docker exec env / systemd show-environment).
Validate the key against https://api.openai.com/v1/models with a curl Bearer request.

Example fix

// before: raises deep in generate()
result = openai_provider.generate(request)
// after: pre-flight config check
if not openai_provider.is_configured():
    raise RuntimeError("OpenAI provider not configured: set external_openai_api_key")
result = openai_provider.generate(request)

Defensive patterns

Strategy: validation

Validate before calling

if not app_config.external_openai_api_key:
    raise RuntimeError("OpenAI is not configured: set external_openai_api_key before using OpenAI models")
# or gate the UI:
if not openai_provider.is_configured():
    hide_openai_models()

Try / catch

try:
    result = provider.generate(request)
except ExternalProviderRequestError as e:
    if "not configured" in str(e):
        show_setup_instructions("openai")
    else:
        raise

Prevention

Call provider.is_configured() before offering OpenAI models in the UI or dispatching jobs.
Ensure the env var is exported inside the actual runtime (docker/systemd), not just your shell.
Add a startup config check that warns when an enabled provider lacks its API key.
Rotate and validate keys periodically with a lightweight authenticated call.

When it happens

Trigger: A generation request routed to provider_id 'openai' while self._app_config.external_openai_api_key is empty/None — including edits requests that would otherwise need reference images. is_configured() likewise returns False.

Common situations: OPENAI_API_KEY env var not exported to the InvokeAI process (e.g. set in shell but not in systemd/docker); key configured under the wrong config attribute; fresh install where only Gemini was configured; container secret not mounted.

Understand the failure class

Background: "environment variable is not set" and "Missing keys in environment" errors: what missing required env var messages mean and how to fix them — this error's family across 28 libraries.

Related errors

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/9c0e7fadaea699a5. Report an issue: GitHub.