BerriAI/litellm · error · ValueError

litellm_logging_obj is required, but got None

Error message

litellm_logging_obj is required, but got None

What it means

In Google GenAI generate_content setup: when no native provider config exists for the model/provider, LiteLLM falls back to the adapter path, which needs a litellm_logging_obj to run the underlying completion call with proper logging. If kwargs passed litellm_logging_obj=None on this fallback branch, setup fails immediately with ValueError.

Source

Thrown at litellm/google_genai/main.py:154

            api_key=litellm_params.api_key,
        )

        if litellm_params.custom_llm_provider is None:
            litellm_params.custom_llm_provider = custom_llm_provider

        # get provider config
        generate_content_provider_config: Final[BaseGoogleGenAIGenerateContentConfig | None] = (
            ProviderConfigManager.get_provider_google_genai_generate_content_config(
                model=model,
                provider=litellm.LlmProviders(custom_llm_provider),
            )
        )

        if generate_content_provider_config is None:
            # Use adapter to transform to completion format when provider config is None
            # Signal that we should use the adapter by returning special result
            if litellm_logging_obj is None:
                raise ValueError("litellm_logging_obj is required, but got None")
            return GenerateContentSetupResult(
                model=model,
                custom_llm_provider=custom_llm_provider,
                request_body={},  # Will be handled by adapter
                generate_content_provider_config=None,
                generate_content_config_dict=dict(config or {}),
                native_request_fields={},
                litellm_params=litellm_params,
                litellm_logging_obj=litellm_logging_obj,
                litellm_call_id=litellm_call_id,
            )

        #########################################################################################
        # Construct request body
        #########################################################################################
        # Create Google Optional Params Config
        generate_content_config_dict: Final = generate_content_provider_config.map_generate_content_optional_params(
            generate_content_config_dict=config or {},

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Call the public generate_content API (litellm.google_genai.generate_content / agenerate_content) which constructs litellm_logging_obj for you
2. If calling internals, build and pass litellm_logging_obj: Logging(model=model, stream=stream, call_type='completion')
3. Ensure custom_llm_provider/model resolve to a provider with a native config if you truly cannot supply a logger

Example fix

# before
setup = generate_content_setup(model='x', contents=c, litellm_logging_obj=None)

# after
from litellm.litellm_core_utils.litellm_logging import Logging
setup = generate_content_setup(model='x', contents=c, litellm_logging_obj=Logging('x', stream=False, call_type='completion'))

if calling_internals:
    from litellm.litellm_core_utils.litellm_logging import Logging
    kwargs["litellm_logging_obj"] = kwargs.get("litellm_logging_obj") or Logging(model, stream=stream, call_type="completion")

def has_valid_logging_obj(lo) -> bool:
    return lo is not None and hasattr(lo, "update_from_kwargs")

When it happens

Trigger: Calling litellm.google_genai generate_content entry points directly without a logging object while the model resolves to a provider lacking a native generate_content config (triggering the adapter branch).

Common situations: Bypassing the public wrapper that creates Logging; custom integrations that call _generate_content_setup-style internals; SDK refactors where the logging kwarg name changed.

Related errors

• logging_obj is required
• logging_obj is required
• logging_obj is required
• Failed to transform streaming response
• Error calling litellm.acompletion for generate_content: {e}

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/aa8e5f1bc9880c26. Report an issue: GitHub.