BerriAI/litellm · error · ValueError

classifier_llm_config is not set

Error message

classifier_llm_config is not set

What it means

Config guard in the complexity router: routing requests need an LLM classifier (classifier_type 'llm' or code paths calling _classify_with_llM) but no classifier_llm_config was set, so there is no model to classify with.

Source

Thrown at litellm/router_strategy/complexity_router/complexity_router.py:1001

        Call the configured classifier model with a system/user role split and prior-turn context.

        Builds a structured classification prompt with:
        - System message: the stable classifier rubric AND the caller's own system prompt (task
          constraints). This is the largest, most repeated part of the call, so keeping it in the
          system role lets the provider prompt-cache it across a session's classifier calls.
        - User message: the variable payload -- a few prior user turns for context and the current
          ask to classify.

        Args:
            prompt: The current user ask text (already extracted as the real human ask, not tool results)
            system_prompt: The caller's system prompt (task constraints), always included so later
                turns never lose it
            request_kwargs: Request metadata for spend attribution
            messages: Full message history for extracting prior turns and the trajectory signal
        """
        llm_config: Final = self.config.classifier_llm_config
        if llm_config is None:
            raise ValueError("classifier_llm_config is not set")

        include_assistant: Final = self.config.classifier_context_include_assistant_turns
        context_enabled: Final = bool(messages) and self.config.classifier_context_window_size > 0
        prior_turns: Final = (
            _extract_prior_turns(
                messages,
                current_ask=prompt,
                window_size=self.config.classifier_context_window_size,
                per_turn_chars=self.config.classifier_context_per_turn_chars,
                include_assistant=include_assistant,
                marker_pairs=self._reminder_markers,
            )
            if context_enabled
            else ()
        )
        has_prior_conversation: Final = (
            context_enabled
            and len(

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. Set classifier_llm_config in the complexity router config.

When it happens

Trigger: Thrown at litellm/router_strategy/complexity_router/complexity_router.py:1001 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/fa0ac7b8cc105df7. Report an issue: GitHub.