BerriAI/litellm · error · ValueError

GEMINI_API_BASE is required

Error message

GEMINI_API_BASE is required

What it means

Endpoint guard in the Gemini ingestion flow: no api_base was resolved from vector_store_config or the GEMINI_API_BASE environment variable, so the target URL for the Gemini File API cannot be constructed. Set GEMINI_API_BASE (e.g. for a proxy or regional endpoint) and retry.

Source

Thrown at litellm/rag/ingestion/gemini_ingestion.py:96

            embeddings: Ignored - Gemini handles embedding
            existing_file_id: Existing provider file ID, unsupported for Gemini

        Returns:
            Tuple of (vector_store_id, file_id)
        """
        vector_store_id = self.vector_store_config.get("vector_store_id")

        vector_store_config: Final = cast(dict[str, Any], self.vector_store_config)

        # Get API credentials
        api_key: Final = cast(str | None, vector_store_config.get("api_key")) or GeminiModelInfo.get_api_key()
        api_base: Final = cast(str | None, vector_store_config.get("api_base")) or GeminiModelInfo.get_api_base()

        if not api_key:
            raise ValueError("GEMINI_API_KEY or GOOGLE_API_KEY is required for Gemini File Search")

        if not api_base:
            raise ValueError("GEMINI_API_BASE is required")

        api_version: Final = "v1beta"
        base_url: Final = f"{api_base}/{api_version}"

        # Create File Search store if not provided
        if not vector_store_id:
            vector_store_id = await self._create_file_search_store(
                api_key=api_key,
                base_url=base_url,
                display_name=self.ingest_name or "litellm-rag-ingest",
            )

        # Upload file to File Search store
        result_file_id = None
        if file_content and filename and vector_store_id:
            result_file_id = await self._upload_to_file_search_store(
                api_key=api_key,
                base_url=base_url,

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. Set GEMINI_API_BASE to the Gemini API base URL.

When it happens

Trigger: Thrown at litellm/rag/ingestion/gemini_ingestion.py:96 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/2fbc92127f80dfa5. Report an issue: GitHub.