BerriAI/litellm · error · ValueError

No API Base provided for Azure OpenAI LLM provider. Set 'AZU

Error message

No API Base provided for Azure OpenAI LLM provider. Set 'AZURE_API_BASE' in .env

What it means

Azure OpenAI embeddings must be sent to your resource's endpoint. litellm resolves it from the api_base kwarg, litellm.api_base, or the AZURE_API_BASE environment variable; when all are unset it raises this ValueError before any network call is made.

Source

Thrown at litellm/main.py:6170

        if azure is True or custom_llm_provider == "azure":
            # azure configs

            api_base = api_base or litellm.api_base or get_secret_str("AZURE_API_BASE")

            api_version = (
                api_version
                or litellm.api_version
                or get_secret_str("AZURE_API_VERSION")
                or litellm.AZURE_DEFAULT_API_VERSION
            )

            azure_ad_token: Final = optional_params.pop("azure_ad_token", None) or get_secret_str("AZURE_AD_TOKEN")

            api_key = api_key or litellm.api_key or litellm.azure_key or get_secret_str("AZURE_API_KEY")

            if api_base is None:
                raise ValueError("No API Base provided for Azure OpenAI LLM provider. Set 'AZURE_API_BASE' in .env")

            ## EMBEDDING CALL
            response = azure_chat_completions.embedding(
                model=model,
                input=input,
                api_base=api_base,
                api_key=api_key,
                api_version=api_version,
                azure_ad_token=azure_ad_token,
                azure_ad_token_provider=azure_ad_token_provider,
                logging_obj=logging,
                timeout=timeout,
                model_response=EmbeddingResponse(),
                optional_params=optional_params,
                client=client,
                aembedding=aembedding,
                max_retries=max_retries,
                headers=headers or extra_headers,

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. export AZURE_API_BASE=https://<your-resource>.openai.azure.com/ and confirm the process sees it
2. Or pass it per call: litellm.embedding(model='azure/<deploy>', input=[...], api_base='https://<resource>.openai.azure.com/')
3. Or set litellm.api_base once if every call targets the same resource
4. Sanity check: python -c "import os; print(os.environ.get('AZURE_API_BASE'))"

Example fix

# before
os.environ["AZURE_API_KEY"] = key
resp = litellm.embedding(model="azure/my-deploy", input=["hi"])

# after
os.environ["AZURE_API_KEY"] = key
os.environ["AZURE_API_BASE"] = "https://my-resource.openai.azure.com/"
resp = litellm.embedding(model="azure/my-deploy", input=["hi"])

import os

api_base = os.environ.get("AZURE_API_BASE") or litellm.api_base
if not api_base:
    raise SystemExit("AZURE_API_BASE is not set; refusing to call azure embeddings")

try:
    resp = litellm.embedding(model="azure/deploy", input=["hi"])
except ValueError as e:
    if "No API Base" in str(e):
        # config problem, not transient: surface to operator
        raise

When it happens

Trigger: litellm.embedding(model='azure/<deployment>', input=[...]) (or custom_llm_provider='azure') with no api_base kwarg, no litellm.api_base, and AZURE_API_BASE absent from the environment.

Common situations: Fresh Azure setup where only AZURE_API_KEY was configured; the variable misnamed AZURE_OPENAI_API_BASE or OPENAI_API_BASE; .env not loaded before import; container/CI dropping env vars.

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

• api_base is required for Azure OpenAI LLM provider. Either s
• Azure AI Agents requests require an api_base. Set `api_base`
• Azure Anthropic requests require an api_base. Set `api_base`
• api_base is required for A2A provider. Either provide api_ba
• api_base is required for triton. Please pass `api_base`

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/b529afcac09759f5. Report an issue: GitHub.