BerriAI/litellm · error · BadRequestError

litellm.BadRequestError

Error message

litellm.BadRequestError: {custom_llm_provider}Exception - {error_str}

What it means

Vertex AI replied 'Vertex AI API has not been used in project' or 'Unable to find your project', and litellm maps it to BadRequestError with a synthetic 400 response and litellm_debug_info. The Google project litellm used either does not exist, is misspelled, or does not have the Vertex AI API (aiplatform.googleapis.com) enabled.

Solutions

1. Enable the API: gcloud services enable aiplatform.googleapis.com --project=<project>
2. Pass the exact project: litellm.completion(..., vertex_project='my-proj', vertex_location='us-east1') and make sure it matches the credentials' project
3. Verify the project id exists as spelled: gcloud projects describe <project>
4. Re-authenticate: gcloud auth application-default login, or point GOOGLE_APPLICATION_CREDENTIALS at a valid service account JSON for that project

Example fix

# before
litellm.completion(
    model='gemini-1.5-pro',
    messages=messages,
    custom_llm_provider='vertex_ai',
)  # project inferred wrongly / API disabled

# after
# 1) gcloud services enable aiplatform.googleapis.com --project=my-proj
litellm.completion(
    model='gemini-1.5-pro',
    messages=messages,
    vertex_project='my-proj',
    vertex_location='us-central1',
)

from googleapiclient import discovery

def vertex_api_enabled(project_id: str, creds) -> bool:
    service = discovery.build('serviceusage', 'v1', credentials=creds)
    state = service.services().get(name=f'projects/{project_id}/services/aiplatform.googleapis.com').execute()
    return state['state'] == 'ENABLED'

# startup check for every project you route to
assert vertex_api_enabled('my-proj', creds)

import litellm

def is_vertex_project_error(exc: BaseException) -> bool:
    msg = str(exc)
    return isinstance(exc, litellm.exceptions.BadRequestError) and (
        'Vertex AI API has not been used in project' in msg or 'Unable to find your project' in msg
    )

import litellm
from litellm.exceptions import BadRequestError

try:
    resp = litellm.completion(model='vertex_ai/gemini-1.5-pro', messages=messages,
                              vertex_project='my-proj', vertex_location='us-central1')
except BadRequestError as e:
    if 'not been used in project' in str(e) or 'Unable to find your project' in str(e):
        raise RuntimeError('enable aiplatform.googleapis.com on project my-proj')
    raise

When it happens

Trigger: custom_llm_provider='vertex_ai' where vertex_project / config project_id is misspelled; the project exists but aiplatform.googleapis.com was never enabled; credentials from project A while the request targets project B.

Common situations: New GCP project where Vertex AI API was never turned on; service account JSON from a different project; vertex_project parameter omitted so litellm resolves the wrong default project.

Related errors

• ContextWindowExceededError
• Exception BadRequestError
• Exception ContentPolicyViolationError
• Exception
• litellm.RateLimitError

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/5ea624c10138879e. Report an issue: GitHub.

Appendix: source

Thrown at litellm/litellm_core_utils/exception_mapping_utils.py:1085

                model=model,
                llm_provider=custom_llm_provider,
                litellm_debug_info=extra_information,
                exception_status_code=original_exception.status_code,
            )


def _map_vertex_exception(
    *,
    model: str,
    original_exception: _ProviderHTTPException,
    custom_llm_provider: str,
    error_str: str,
    exception_type: str,
    exception_provider: str,
    extra_information: str,
) -> None:
    if "Vertex AI API has not been used in project" in error_str or "Unable to find your project" in error_str:
        raise BadRequestError(
            message=f"litellm.BadRequestError: {custom_llm_provider}Exception - {error_str}",
            model=model,
            llm_provider=custom_llm_provider,
            response=httpx.Response(
                status_code=400,
                request=httpx.Request(
                    method="POST",
                    url=" https://cloud.google.com/vertex-ai/",
                ),
            ),
            litellm_debug_info=extra_information,
        )
    if "400 Request payload size exceeds" in error_str:
        raise ContextWindowExceededError(
            message=f"{custom_llm_provider.capitalize()}Exception - {error_str}",
            model=model,
            llm_provider=custom_llm_provider,
        )

View on GitHub (pinned to 77b7c6c40c)