BerriAI/litellm · error · ValueError

Reducto expected OCR preprocessing to produce document_url o

Error message

Reducto expected OCR preprocessing to produce document_url or image_url for model={model}

What it means

Preprocessing guard in the Reducto OCR transformation: the supplied DocumentType contains neither document_url nor image_url, so there is no source to send to the /parse endpoint and the request is rejected before URL construction.

Source

Thrown at litellm/llms/reducto/ocr/transformation.py:71

            "Authorization": f"Bearer {resolved_key}",
            "Content-Type": "application/json",
            **headers,
        }

    def get_complete_url(
        self,
        api_base: str | None,
        model: str,
        optional_params: dict,
        litellm_params: dict | None = None,
        **kwargs,
    ) -> str:
        return "{}/parse".format((api_base or REDUCTO_API_BASE).rstrip("/"))

    def _get_source_url(self, document: DocumentType, model: str) -> str:
        source_url: Final = document.get("document_url") or document.get("image_url")
        if source_url is None:
            raise ValueError(
                f"Reducto expected OCR preprocessing to produce document_url or image_url for model={model}"
            )
        return source_url

    @staticmethod
    def _resolve_credentials(api_key: str | None, api_base: str | None) -> tuple[str, str]:
        from litellm.secret_managers.main import get_secret_str

        resolved_key: Final = api_key or get_secret_str("REDUCTO_API_KEY")
        if resolved_key is None:
            raise ValueError(
                "Missing REDUCTO_API_KEY - set it in the environment or pass api_key to litellm.ocr()/litellm.aocr()"
            )
        resolved_base: Final = (api_base or REDUCTO_API_BASE).rstrip("/")
        return resolved_key, resolved_base

    def _ensure_file_id_sync(
        self,

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. Ensure the OCR preprocessing step yields a document_url or image_url for the chosen model.
2. Pass a supported input (document or image) for the specified Reducto model.

Example fix

# provide a file or URL that preprocessing converts to document_url/image_url.

When it happens

Trigger: Triggered when Reducto OCR preprocessing fails to produce a document_url or image_url for the requested model.

Common situations: See trigger scenarios.

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/9477529bc23189b3. Report an issue: GitHub.