BerriAI/litellm · error · ValueError

custom_llm_provider is required

Error message

custom_llm_provider is required

What it means

Thrown by SpeechToCompletionBridgeHandler.validate_input_kwargs when 'custom_llm_provider' is absent from kwargs or is not a str (handler.py:41). The bridge needs the resolved provider name (e.g. 'openai') to transform the TTS request into a chat/completions call. Because litellm's main_router normally injects custom_llm_provider, hitting this means the call bypassed standard routing or the provider could not be resolved.

Source

Thrown at litellm/endpoints/speech/speech_to_completion_bridge/handler.py:41

class SpeechToCompletionBridgeHandler:
    def __init__(self):
        from .transformation import SpeechToCompletionBridgeTransformationHandler

        super().__init__()
        self.transformation_handler = SpeechToCompletionBridgeTransformationHandler()

    def validate_input_kwargs(self, kwargs: dict) -> SpeechToCompletionBridgeHandlerInputKwargs:
        from litellm import LiteLLMLoggingObj

        model: Final = kwargs.get("model")
        if model is None or not isinstance(model, str):
            raise ValueError("model is required")

        custom_llm_provider: Final = kwargs.get("custom_llm_provider")
        if custom_llm_provider is None or not isinstance(custom_llm_provider, str):
            raise ValueError("custom_llm_provider is required")

        input: Final = kwargs.get("input")
        if input is None or not isinstance(input, str):
            raise ValueError("input is required")

        optional_params: Final = kwargs.get("optional_params")
        if optional_params is None or not isinstance(optional_params, dict):
            raise ValueError("optional_params is required")

        litellm_params: Final = kwargs.get("litellm_params")
        if litellm_params is None or not isinstance(litellm_params, dict):
            raise ValueError("litellm_params is required")

        headers = kwargs.get("headers")
        if headers is None or not isinstance(headers, dict):
            raise ValueError("headers is required")

        headers = kwargs.get("headers")

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Use litellm.audio_speech()/litellm.speech() rather than invoking the bridge handler directly, so routing populates custom_llm_provider
2. If direct invocation is required, pass custom_llm_provider="openai" explicitly
3. Prefix the model with the provider (e.g. 'openai/gpt-4o-audio-preview') so the provider is resolvable

Example fix

# before
speech_to_completion_bridge_handler.speech(model="gpt-4o-audio-preview", input="hi", voice="alloy", optional_params={}, litellm_params={}, headers={}, logging_obj=logging_obj)

# after
speech_to_completion_bridge_handler.speech(model="gpt-4o-audio-preview", input="hi", voice="alloy", optional_params={}, litellm_params={}, headers={}, logging_obj=logging_obj, custom_llm_provider="openai")

provider = kwargs.get("custom_llm_provider")
if not isinstance(provider, str):
    # let litellm routing resolve it instead of calling the handler directly
    kwargs["model"] = f"openai/{model}"  # explicit provider prefix
    # or: kwargs["custom_llm_provider"] = "openai"

def has_resolved_provider(kwargs: dict) -> bool:
    return isinstance(kwargs.get("custom_llm_provider"), str)

try:
    resp = litellm.audio_speech(model="openai/gpt-4o-audio-preview", input=text, voice=voice)
except ValueError as e:
    if "custom_llm_provider is required" in str(e):
        logger.error("Provider unresolved; prefix model with provider name")
        raise

When it happens

Trigger: Calling speech_to_completion_bridge_handler.speech(...) directly without custom_llm_provider; a model string formatted so get_llm_provider cannot resolve a provider (e.g. an unknown prefix) causing custom_llm_provider to be dropped; patched/mocked routing code that strips the key from kwargs.

Common situations: Custom integrations that call the bridge handler instead of litellm.audio_speech; model names without a recognizable provider prefix ('openai/...', 'azure/...'); unit tests with mocked router outputs.

Related errors

• litellm_params is required
• prompt_characters must be provided for tts calls. prompt_cha
• model is required
• input is required
• optional_params is required

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/b1623ea81261152f. Report an issue: GitHub.