BerriAI/litellm · error · ValueError

api_base is required for LangGraph. Set it via LANGGRAPH_API

Error message

api_base is required for LangGraph. Set it via LANGGRAPH_API_BASE env var or api_base parameter.

What it means

The LangGraph provider builds its endpoint as {api_base}/runs/stream (streaming) or {api_base}/runs/wait (non-streaming) and has no default host. get_complete_url raises ValueError when api_base is None — it must be supplied via the api_base parameter or the LANGGRAPH_API_BASE environment variable (typically the LangGraph Platform/Agent Server URL).

Source

Thrown at litellm/llms/langgraph/chat/transformation.py:106

        return optional_params

    def get_complete_url(
        self,
        api_base: str | None,
        api_key: str | None,
        model: str,
        optional_params: dict,
        litellm_params: dict,
        stream: bool | None = None,
    ) -> str:
        """
        Get the complete URL for the LangGraph request.

        Streaming: /runs/stream
        Non-streaming: /runs/wait
        """
        if api_base is None:
            raise ValueError(
                "api_base is required for LangGraph. Set it via LANGGRAPH_API_BASE env var or api_base parameter."
            )

        # Remove trailing slash if present
        api_base = api_base.rstrip("/")

        # Choose endpoint based on streaming mode
        if stream:
            return f"{api_base}/runs/stream"
        else:
            return f"{api_base}/runs/wait"

    def _get_assistant_id(self, model: str, optional_params: dict) -> str:
        """
        Get the assistant ID from model or optional_params.

        model format: "langgraph/assistant_id" or just "assistant_id"
        """

View on GitHub (pinned to 6c2dcb801b)

Solutions

Set LANGGRAPH_API_BASE to your LangGraph server URL (e.g. https://my-agent.langgraph.com)
Or pass api_base=... explicitly on the call / model_list entry
For LiteLLM proxy, add api_base to the deployment's litellm_params for the langgraph model
Confirm the env var is actually present in the process: printenv LANGGRAPH_API_BASE

Example fix

# before
litellm.completion(model="langgraph/my-agent", messages=msgs, api_key=lg_key)

# after
litellm.completion(
    model="langgraph/my-agent",
    messages=msgs,
    api_key=lg_key,
    api_base="https://my-agent.langgraph.com",
)

Defensive patterns

Strategy: validation

Validate before calling

import os

def ensure_langgraph_config(api_base: str | None = None) -> str:
    base = api_base or os.getenv("LANGGRAPH_API_BASE")
    if not base:
        raise ValueError("LANGGRAPH_API_BASE is not set; refusing to call LangGraph")
    return base.rstrip("/")

Try / catch

try:
    litellm.completion(model="langgraph/agent", messages=msgs, api_base=ensure_langgraph_config())
except ValueError as e:
    if "api_base is required for LangGraph" in str(e):
        raise RuntimeError("Bootstrap error: LANGGRAPH_API_BASE missing in this environment") from e

Prevention

Pin LANGGRAPH_API_BASE (and LANGGRAPH_API_KEY) in the same env manifest for every environment
Add a startup connectivity check against {base}/info or /runs before serving traffic
Watch for the common typo LANGGRAPH_API_URL — assert the exact variable name in tests

When it happens

Trigger: Calling completion(model="langgraph/...", stream=True) with neither api_base nor LANGGRAPH_API_BASE set; env var defined only in the local shell but not in the deployed container/CI.

Common situations: Missing LANGGRAPH_API_BASE when integrating a LangGraph Deployment (e.g. https://my-deployment.langgraph.com); typo'd variable name (LANGGRAPH_API_URL); .env not loaded; proxy config lacking api_base on the langgraph model entry.

Related errors

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/8c777f19824c8e13. Report an issue: GitHub.