What causes "'models' param not in kwargs" in BerriAI/litellm?

BerriAI/litellm · error · Exception

'models' param not in kwargs

Error message

'models' param not in kwargs

What it means

Generic Exception raised by litellm.batch_completion_models when the kwargs dict does not contain a 'models' key. The function is designed to be called as batch_completion_models(**kwargs) with 'models' supplied among the kwargs; it pops 'models' (and 'model') out and fans the rest out to litellm.completion per model. Without 'models' there is nothing to parallelize.

Source

Thrown at litellm/batch_completion/main.py:230

            - Other keyword arguments to be passed to the completion function.

    Returns:
        list: A list of responses from the language models that responded.

    Note:
        This function utilizes a ThreadPoolExecutor to parallelize requests to multiple models.
        It sends requests concurrently and collects responses from all models that respond.
    """
    import concurrent.futures

    # ANSI escape codes for colored output

    if "model" in kwargs:
        kwargs.pop("model")
    if "models" in kwargs:
        models = kwargs.pop("models")
    else:
        raise Exception("'models' param not in kwargs")

    if isinstance(models, str):
        models = [models]
    elif isinstance(models, (list, tuple)):
        models = list(models)
    else:
        raise TypeError("'models' must be a string or list of strings")

    if len(models) == 0:
        return []

    responses: Final = []

    with concurrent.futures.ThreadPoolExecutor(max_workers=len(models)) as executor:
        futures: Final = [executor.submit(litellm.completion, *args, model=model, **kwargs) for model in models]

        for future in futures:
            try:

View on GitHub (pinned to 6c2dcb801b)

Solutions

Call it as litellm.batch_completion_models(models=['gpt-4o', 'claude-3-5-sonnet'], messages=[...]).
If you meant many prompts for one model, use litellm.batch_completion(models='gpt-4o', messages=[...]) instead.
Ensure 'models' is a keyword argument, not positional.

Example fix

# before
litellm.batch_completion_models(["gpt-4o", "claude-3-5-sonnet"], messages=[{"role": "user", "content": "hi"}])

# after
litellm.batch_completion_models(models=["gpt-4o", "claude-3-5-sonnet"], messages=[{"role": "user", "content": "hi"}])

Defensive patterns

Strategy: validation

Validate before calling

if "models" not in kwargs or not kwargs["models"]:
    raise ValueError("batch_completion_models requires a non-empty 'models' kwarg")
kwargs.setdefault("models", list(kwargs["models"]))

Type guard

def is_valid_batch_models_arg(kwargs: dict) -> bool:
    return isinstance(kwargs.get("models"), (str, list, tuple)) and len(kwargs.get("models", ())) or isinstance(kwargs.get("models"), str)

Prevention

Always call batch_completion_models(models=..., ...) with models as a keyword.
Use batch_completion (singular model, many prompts) when you have one model.
Add a wrapper function with an explicit models: list[str] parameter to get static checking.

When it happens

Trigger: Calling litellm.batch_completion_models() without a models kwarg, e.g. passing models as a positional arg (unsupported) or forgetting it entirely: litellm.batch_completion_models(prompt='hi'). Note 'model' is silently popped first, so passing only model=... does not help.

Common situations: Confusing batch_completion_models (list of models, one prompt) with batch_completions (list of prompts); passing models positionally because the signature accepts *args for completion args; refactoring that drops the kwarg.

Related errors

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/fc5288be39e955c2. Report an issue: GitHub.