BerriAI/litellm · error · ValueError

Unable to get vertex url/endpoint for mode: {mode}

Error message

Unable to get vertex url/endpoint for mode: {mode}

What it means

Vertex URL mapper sentinel: get_url for the given mode (e.g. an unsupported operation type) has no URL-building branch, so it raises to signal the mode/path combination is unhandled rather than returning a bogus endpoint.

Source

Thrown at litellm/llms/vertex_ai/common_utils.py:446

            vertex_project=vertex_project,
            vertex_location=vertex_location,
            vertex_api_version=vertex_api_version,
        )
    elif mode == "image_generation":
        endpoint = "predict"
        base_url = get_vertex_base_url(vertex_location)
        if model.isdigit():
            # Numeric model -> custom endpoint
            url = f"{base_url}/{vertex_api_version}/projects/{vertex_project}/locations/{vertex_location}/endpoints/{model}:{endpoint}"
        else:
            # Regular model -> publisher model
            url = f"{base_url}/v1/projects/{vertex_project}/locations/{vertex_location}/publishers/google/models/{model}:{endpoint}"
    elif mode == "count_tokens":
        endpoint = "countTokens"
        base_url = get_vertex_base_url(vertex_location)
        url = f"{base_url}/{vertex_api_version}/projects/{vertex_project}/locations/{vertex_location}/publishers/google/models/{model}:{endpoint}"
    if not url or not endpoint:
        raise ValueError(f"Unable to get vertex url/endpoint for mode: {mode}")
    return url, endpoint


def _get_gemini_url(
    mode: all_gemini_url_modes,
    model: str,
    stream: bool | None,
) -> tuple[str, str]:
    """Build the Gemini API URL for the given mode.

    The API key is NOT included in the URL. Callers must pass it via the
    ``x-goog-api-key`` header instead to avoid leaking credentials in
    error tracebacks.
    """
    from litellm.llms.vertex_ai.gemini.vertex_and_google_ai_studio_gemini import (
        VertexGeminiConfig,
    )

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. Use a supported mode for Vertex AI (chat, embedding, image_generation, etc.).
2. Check the call type and provider route configuration.

Example fix

# verify the litellm call function matches a supported vertex mode.

When it happens

Trigger: Triggered when no Vertex AI URL/endpoint can be constructed for the requested mode.

Common situations: See trigger scenarios.

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/cc781e0059db9954. Report an issue: GitHub.