microsoft/semantic-kernel · error · ServiceInitializationError

The DeepSeek API key is required.

Error message

The DeepSeek API key is required.

What it means

ServiceInitializationError raised by the DeepSeek setup helper when OpenAISettings (reading env/credential store) has no api_key. DeepSeek is wired through the OpenAI client pointed at api.deepseek.com, so it needs a valid API key; failing fast at construction prevents a later, harder-to-diagnose 401 at request time. The exception type signals this is a service-init problem, not a transient runtime error.

Source

Thrown at python/samples/concepts/setup/chat_completion_services.py:394

    Set the `OPENAI_CHAT_MODEL_ID` environment variable to the DeepSeek model ID (deepseek-chat or deepseek-reasoner).

    The request settings control the behavior of the service. The default settings are sufficient to get started.
    However, you can adjust the settings to suit your needs.
    Note: Some of the settings are NOT meant to be set by the user.
    Please refer to the Semantic Kernel Python documentation for more information:
    https://learn.microsoft.com/en-us/python/api/semantic-kernel/semantic_kernel?view=semantic-kernel-python
    """
    from openai import AsyncOpenAI

    from semantic_kernel.connectors.ai.open_ai import (
        OpenAIChatCompletion,
        OpenAIChatPromptExecutionSettings,
        OpenAISettings,
    )

    openai_settings = OpenAISettings()
    if not openai_settings.api_key:
        raise ServiceInitializationError("The DeepSeek API key is required.")
    if not openai_settings.chat_model_id:
        raise ServiceInitializationError("The DeepSeek model ID is required.")

    chat_service = OpenAIChatCompletion(
        ai_model_id=openai_settings.chat_model_id,
        service_id=service_id,
        async_client=AsyncOpenAI(
            api_key=openai_settings.api_key.get_secret_value(),
            base_url="https://api.deepseek.com",
        ),
    )
    request_settings = OpenAIChatPromptExecutionSettings(service_id=service_id)

    return chat_service, request_settings


def get_nvidia_chat_completion_service_and_request_settings() -> tuple[
    "ChatCompletionClientBase", "PromptExecutionSettings"

View on GitHub (pinned to c028a0c7dc)

Solutions

1. Set the DeepSeek API key in the environment OpenAISettings reads (e.g. export DEEPSEEK_API_KEY=... or add it to your .env that the sample loads).
2. Verify with a quick check before construction: print whether the env var is set (name only, never the value).
3. Use a secrets manager / devkey flow to inject the key rather than hard-coding it.
4. Confirm OpenAISettings is reading the right variable name for your configuration.

Example fix

# before
openai_settings = OpenAISettings()
if not openai_settings.api_key:
    raise ServiceInitializationError("The DeepSeek API key is required.")
# after - fail with an actionable message pointing at the env var
openai_settings = OpenAISettings()
if not openai_settings.api_key:
    raise ServiceInitializationError(
        "The DeepSeek API key is required. Set DEEPSEEK_API_KEY in your environment."
    )

import os
from semantic_kernel.connectors.ai.open_ai import OpenAISettings

def ensure_deepseek_key():
    settings = OpenAISettings()
    if not settings.api_key:
        raise ServiceInitializationError(
            "Set DEEPSEEK_API_KEY in your environment before constructing the service."
        )
    return settings

def has_deepseek_key() -> bool:
    settings = OpenAISettings()
    return bool(settings.api_key)

try:
    service, settings = get_deepseek_chat_completion_service_and_request_settings()
except ServiceInitializationError as e:
    if "API key" in str(e):
        # prompt the user/secrets flow to provide DEEPSEEK_API_KEY, then retry
        ...
    raise

When it happens

Trigger: Calling get_deepseek_chat_completion_service_and_request_settings() when the DEEPSEEK_API_KEY (or whichever env var OpenAISettings binds to) is unset/empty, so openai_settings.api_key is falsy.

Common situations: Missing environment variable for the DeepSeek key; the .env file not loaded; the key set under a different variable name than OpenAISettings expects; running on a machine/CI without the secret; or a copy-paste that set the model id but not the key.

Related errors

• The vector store must have an embedding generator.
• Unsupported service name: {service_name}
• Failed to validate Azure AI Inference settings: {e}
• Failed to initialize the Amazon Bedrock Chat Completion Serv
• The Amazon Bedrock Chat Model ID is missing.

AI-assisted analysis of microsoft/semantic-kernel@c028a0c7dc (2026-08-13). Data as JSON: /api/errors/f953b03e035efc8e. Report an issue: GitHub.