langchain-ai/deepagents · error · ModelConfigError

Google Cloud location is required for provider 'google_anthr

Error message

Google Cloud location is required for provider 'google_anthropic_vertex'. Set GOOGLE_CLOUD_LOCATION or DEEPAGENTS_CODE_GOOGLE_CLOUD_LOCATION, or pass 'location' in model params.

What it means

Raised when configuring the `google_anthropic_vertex` provider (Claude on Vertex AI) without a Google Cloud location. Vertex requires a project and a region; the library injects them from stored credentials or config params via `_apply_google_anthropic_vertex_kwargs` and refuses to guess the region. GCP endpoints are region-specific, so there is no safe default.

Source

Thrown at libs/code/deepagents_code/config.py:5309

    Raises:
        ModelConfigError: If no location is configured.
    """
    if provider != "google_anthropic_vertex":
        return
    credentials = _get_credentials()
    if credentials.google_cloud_project:
        kwargs.setdefault("project", credentials.google_cloud_project)
    if credentials.google_cloud_location:
        kwargs.setdefault("location", credentials.google_cloud_location)
    if not kwargs.get("location"):
        from deepagents_code.model_config import ModelConfigError

        msg = (
            "Google Cloud location is required for provider "
            "'google_anthropic_vertex'. Set GOOGLE_CLOUD_LOCATION or "
            "DEEPAGENTS_CODE_GOOGLE_CLOUD_LOCATION, or pass 'location' in model params."
        )
        raise ModelConfigError(msg)


def _compose_openai_reasoning_effort(
    provider: str,
    kwargs: dict[str, Any],
    effort_override: object,
    reasoning_override: object,
) -> dict[str, Any]:
    """Compose a session effort override with an OpenAI reasoning mapping.

    Args:
        provider: Resolved model provider.
        kwargs: Layered model constructor parameters.
        effort_override: High-priority `reasoning_effort` from session params.
        reasoning_override: High-priority native `reasoning` from session params.

    Returns:
        Constructor parameters with one native `reasoning` mapping when

View on GitHub (pinned to a1af029e6e)

Solutions

Set `export GOOGLE_CLOUD_LOCATION=us-east5` (or your region)
Set DEEPAGENTS_CODE_GOOGLE_CLOUD_LOCATION as the app-specific override
Pass `location` in model params: `[providers."google_anthropic_vertex".params] location = "us-east5"` in config.toml or as a model kwarg

Example fix

// before
$ dcode -m google_anthropic_vertex:claude-sonnet-4
ModelConfigError: Google Cloud location is required...

// after
$ export GOOGLE_CLOUD_PROJECT=my-project
$ export GOOGLE_CLOUD_LOCATION=us-east5
$ dcode -m google_anthropic_vertex:claude-sonnet-4

Defensive patterns

Strategy: validation

Validate before calling

import os
def vertex_location_ok() -> bool:
    return bool(
        os.environ.get("GOOGLE_CLOUD_LOCATION")
        or os.environ.get("DEEPAGENTS_CODE_GOOGLE_CLOUD_LOCATION")
    )

Try / catch

from deepagents_code.model_config import ModelConfigError
try:
    model = create_model("google_anthropic_vertex:claude-sonnet-4")
except ModelConfigError as e:
    if "location is required" in str(e):
        os.environ.setdefault("GOOGLE_CLOUD_LOCATION", "us-east5")
        model = create_model("google_anthropic_vertex:claude-sonnet-4")
    else:
        raise

Prevention

Always set GOOGLE_CLOUD_PROJECT and GOOGLE_CLOUD_LOCATION together for Vertex
Put `location` in the provider's `params` table in config.toml for team-wide configs
Match the region to one where the Claude models are available

When it happens

Trigger: Creating a model with provider `google_anthropic_vertex` when, after merging stored credentials and config.toml `params`, `kwargs` still has no `location`: GOOGLE_CLOUD_LOCATION and DEEPAGENTS_CODE_GOOGLE_CLOUD_LOCATION are unset and no `location` appears in model params (config.py:5294-5309).

Common situations: Authenticating via Application Default Credentials where the project is detected but the region is not; CI environments with only GOOGLE_CLOUD_PROJECT set; copying a working config from another region without mirroring the location.

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

AI-assisted analysis of langchain-ai/deepagents@a1af029e6e (2026-08-29). Data as JSON: /api/errors/f21ddec2eb10fc2c. Report an issue: GitHub.