BerriAI/litellm · error · OpenAIError

Missing model or messages

Error message

Missing model or messages

What it means

The OpenAI text-completion handler validates that both `model` and `messages` are provided before building the request. If either is None, it raises OpenAIError 422 'Missing model or messages' immediately — this handler is the legacy /v1/completions text path, which still requires the messages/prompt source to be present.

Source

Thrown at litellm/llms/openai/completion/handler.py:58

        custom_llm_provider: str,
        logging_obj: LiteLLMLoggingObj,
        optional_params: dict,
        print_verbose: Callable | None = None,
        api_base: str | None = None,
        acompletion: bool = False,
        litellm_params=None,
        logger_fn=None,
        client=None,
        organization: str | None = None,
        headers: dict | None = None,
    ):
        try:
            if headers:
                optional_params = {**optional_params, "extra_headers": headers}
            if headers is None:
                headers = self.validate_environment(api_key=api_key)
            if model is None or messages is None:
                raise OpenAIError(status_code=422, message="Missing model or messages")

            # don't send max retries to the api, if set

            provider_config: Final = ProviderConfigManager.get_provider_text_completion_config(
                model=model,
                provider=LlmProviders(custom_llm_provider),
            )

            data: Final = provider_config.transform_text_completion_request(
                model=model,
                messages=messages,
                optional_params=optional_params,
                headers=headers,
            )
            max_retries: Final = data.pop("max_retries", 2)
            ## LOGGING
            logging_obj.pre_call(
                input=messages,

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Check the calling code: ensure the model string resolved (not None) and messages are passed.
2. Default model from env: `model = model or os.environ['OPENAI_MODEL']` with explicit failure if unset.
3. If you meant to send a raw prompt, use the appropriate text-completion parameter form rather than messages=None.

Example fix

# before
litellm.text_completion(model=None, prompt="hi")  # model never resolved

# after
model = os.environ.get("OPENAI_MODEL")
if model is None:
    raise ValueError("OPENAI_MODEL not set")
litellm.text_completion(model=model, prompt="hi")

if not model or not messages:
    raise ValueError(f"model and messages are required (got model={model!r})")

def has_model_and_messages(model, messages) -> bool:
    return isinstance(model, str) and bool(model) and messages is not None

When it happens

Trigger: Calling the text-completion entrypoint (openai text completion, e.g. via `litellm.completion` internals or provider text-completion routes) with model=None (e.g. a variable that failed to resolve) or messages=None.

Common situations: Programmatic callers building params dynamically where the model name variable is None (unset env/config), or wrappers that pass prompt strings on a path that expects messages. Almost always a caller bug, not a server issue.

Related errors

• Braintrust API error: {e.response.text}
• Failed to connect to Braintrust API: {str(e)}
• Failed to parse Braintrust API response: {str(e)}
• error_text
• Unclassified keys in {PRICES_PATH.name}: {', '.join(unclassi

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/8d84c5b835041944. Report an issue: GitHub.