microsoft/semantic-kernel · error · ServiceInitializationError

Failed to create NVIDIA settings.

Error message

Failed to create NVIDIA settings.

What it means

Raised as ServiceInitializationError when constructing the pydantic `NvidiaSettings` for the NVIDIA text embedding service raises a `ValidationError`. Config-only failure, not network. Note: this connector is marked @experimental and does NOT hard-require an api_key (it only warns), so the most common trigger is an invalid base_url or bad env_file_path, not a missing key.

Source

Thrown at python/semantic_kernel/connectors/ai/nvidia/services/nvidia_text_embedding.py:69

                (Env var NVIDIA_API_KEY)
            base_url: HttpsUrl | None - base_url: The url of the NVIDIA endpoint. The base_url consists of the endpoint,
                and more information refer https://docs.api.nvidia.com/nim/reference/
                use endpoint if you only want to supply the endpoint.
                (Env var NVIDIA_BASE_URL)
            client (Optional[AsyncOpenAI]): An existing client to use. (Optional)
            env_file_path (str | None): Use the environment settings file as
                a fallback to environment variables. (Optional)
            service_id (str): Service ID for the model. (optional)
        """
        try:
            nvidia_settings = NvidiaSettings(
                api_key=api_key,
                base_url=base_url,
                embedding_model_id=ai_model_id,
                env_file_path=env_file_path,
            )
        except ValidationError as ex:
            raise ServiceInitializationError("Failed to create NVIDIA settings.", ex) from ex
        if not nvidia_settings.embedding_model_id:
            nvidia_settings.embedding_model_id = "nvidia/nv-embedqa-e5-v5"
            logger.warning(f"Default embedding model set as: {nvidia_settings.embedding_model_id}")
        if not nvidia_settings.api_key:
            logger.warning("API_KEY is missing, inference may fail.")
        if not client:
            client = AsyncOpenAI(
                api_key=nvidia_settings.api_key.get_secret_value() if nvidia_settings.api_key else None,
                base_url=nvidia_settings.base_url,
            )
        super().__init__(
            ai_model_id=nvidia_settings.embedding_model_id,
            api_key=nvidia_settings.api_key.get_secret_value() if nvidia_settings.api_key else None,
            ai_model_type=NvidiaModelTypes.EMBEDDING,
            service_id=service_id or nvidia_settings.embedding_model_id,
            env_file_path=env_file_path,
            client=client,
        )

View on GitHub (pinned to c028a0c7dc)

Solutions

1. Inspect the chained `ex` ValidationError for the failing field.
2. Pass a valid base_url (full https:// URL) and/or env_file_path that exists.
3. Set NVIDIA_API_KEY/NVIDIA_BASE_URL/NVIDIA_EMBEDDING_MODEL_ID correctly in env/.env.
4. If you only need defaults, omit ai_model_id so the built-in default applies after validation.

Example fix

# before
svc = NvidiaTextEmbedding()  # -> ServiceInitializationError

# after
svc = NvidiaTextEmbedding(
    api_key=os.environ['NVIDIA_API_KEY'],
    base_url='https://integrate.api.nvidia.com/v1',
)

from semantic_kernel.connectors.ai.nvidia import NvidiaSettings
from pydantic import ValidationError
try:
    NvidiaSettings(api_key=os.environ.get('NVIDIA_API_KEY'),
                   base_url=os.environ.get('NVIDIA_BASE_URL', 'https://integrate.api.nvidia.com/v1'),
                   embedding_model_id=os.environ.get('NVIDIA_EMBEDDING_MODEL_ID'))
except ValidationError as e:
    raise SystemExit(f'Fix NVIDIA settings first: {e}')

from semantic_kernel.exceptions import ServiceInitializationError

def is_nvidia_embed_settings_error(e: BaseException) -> bool:
    return isinstance(e, ServiceInitializationError) and 'Failed to create NVIDIA settings' in str(e)

from semantic_kernel.exceptions import ServiceInitializationError
try:
    svc = NvidiaTextEmbedding()
except ServiceInitializationError as e:
    raise SystemExit(f'NVIDIA embedding misconfigured: {e.__cause__ or e}') from e

When it happens

Trigger: Constructing `NvidiaTextEmbedding(...)` where merged config fails pydantic NvidiaSettings validation: malformed NVIDIA_BASE_URL (not a valid HttpUrl), unreadable env_file_path, or embedding_model_id set to an invalid type. The default 'nvidia/nv-embedqa-e5-v5' applies only AFTER settings build, so a validation failure short-circuits it.

Common situations: NVIDIA_BASE_URL missing scheme, env_file_path pointing at a non-existent file, a pydantic version difference, or a non-string passed for ai_model_id.

Related errors

• Failed to validate Mistral AI settings: {e}
• Failed to create NVIDIA settings.
• Failed to validate Google AI settings: {e}
• Failed to create Ollama settings.
• Failed to create Ollama settings.

AI-assisted analysis of microsoft/semantic-kernel@c028a0c7dc (2026-08-13). Data as JSON: /api/errors/856e2ddb64e11be3. Report an issue: GitHub.