BerriAI/litellm · error · UnauthorizedError

{e}

Error message

{e}

What it means

Raised by ModelGroupsManagementClient.info() when GET {base_url}/model_group/info returns HTTP 401, mapped to UnauthorizedError (redacted message; original HTTPError on .orig_exception; any other non-2xx re-raises as plain requests HTTPError). On success the method returns response.json()["data"], so authentication is the only failure mode it translates into a library-specific exception.

Source

Thrown at litellm/proxy/client/model_groups.py:61

        Raises:
            UnauthorizedError: If the request fails with a 401 status code
            requests.exceptions.RequestException: If the request fails with any other error
        """
        url: Final = f"{self._base_url}/model_group/info"
        request: Final = requests.Request("GET", url, headers=self._get_headers())

        if return_request:
            return request

        # Prepare and send the request
        session: Final = requests.Session()
        try:
            response: Final = session.send(request.prepare())
            response.raise_for_status()
            return response.json()["data"]
        except requests.exceptions.HTTPError as e:
            if e.response.status_code == 401:
                raise UnauthorizedError(e)
            raise

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. Provide the admin key: ModelGroupsManagementClient(base_url, api_key=os.environ["LITELLM_MASTER_KEY"]).info()
2. Validate the key with curl -H "Authorization: Bearer $KEY" $BASE_URL/model_group/info
3. Refresh credentials after rotation and confirm base_url points at the right proxy

Example fix

# before
mg = ModelGroupsManagementClient("http://localhost:4000")
mg.info()  # UnauthorizedError

# after
import os
mg = ModelGroupsManagementClient("http://localhost:4000", api_key=os.environ["LITELLM_MASTER_KEY"])
mg.info()

import requests

def model_groups_readable(base_url: str, api_key: str | None) -> bool:
    if not api_key:
        return False
    r = requests.get(
        f"{base_url.rstrip('/')}/model_group/info",
        headers={"Authorization": f"Bearer {api_key}"},
        timeout=10,
    )
    return r.status_code != 401

from litellm.proxy.client.exceptions import UnauthorizedError
import requests

try:
    groups = model_groups.info()
except UnauthorizedError:
    rotate_admin_key()  # 401 — credentials rejected
except requests.exceptions.HTTPError as e:
    log_and_surface(e.response)  # any other non-2xx

When it happens

Trigger: Calling model_groups.info() with no api_key on an auth-enforcing proxy; using a virtual/end-user key for what is a management endpoint; a rotated or revoked admin key.

Common situations: Monitoring tools scraping model-group spend/tps data with stale credentials; proxy upgraded and master key changed; script moved between environments without updating the key.

Related errors

• {e}
• {e}
• {e}
• Authentication failed. Check your Arize Phoenix API key and
• Authentication failed. Check your BitBucket access token and

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/cab13ca9c4c4cf28. Report an issue: GitHub.