langchain-ai/langchain · error · ValueError

metadata must be a list of the same length as prompts

Error message

metadata must be a list of the same length as prompts

What it means

Raised by BaseLLM.generate during batch generation with per-prompt callbacks. If metadata is provided alongside a per-prompt callbacks list, metadata must be a list of dicts with one dict per prompt. A single dict or a mismatched-length list is rejected.

Source

Thrown at libs/core/langchain_core/language_models/llms.py:949

            and (
                isinstance(callbacks[0], (list, BaseCallbackManager))
                or callbacks[0] is None
            )
        ):
            # We've received a list of callbacks args to apply to each input
            if len(callbacks) != len(prompts):
                msg = "callbacks must be the same length as prompts"
                raise ValueError(msg)
            if tags is not None and not (
                isinstance(tags, list) and len(tags) == len(prompts)
            ):
                msg = "tags must be a list of the same length as prompts"
                raise ValueError(msg)
            if metadata is not None and not (
                isinstance(metadata, list) and len(metadata) == len(prompts)
            ):
                msg = "metadata must be a list of the same length as prompts"
                raise ValueError(msg)
            if run_name is not None and not (
                isinstance(run_name, list) and len(run_name) == len(prompts)
            ):
                msg = "run_name must be a list of the same length as prompts"
                raise ValueError(msg)
            tags_list = cast("list[list[str] | None]", tags or ([None] * len(prompts)))
            metadata_list = cast(
                "list[builtins.dict[str, Any] | None]",
                metadata or ([{}] * len(prompts)),
            )
            run_name_list = run_name or cast(
                "list[str | None]", ([None] * len(prompts))
            )
            params = self._dict_for_compat()
            params["stop"] = stop
            callback_managers = [
                CallbackManager.configure(
                    callback,

View on GitHub (pinned to e32fa9a52e)

Solutions

1. Pass metadata as a list of dicts of length len(prompts): metadata=[md] * len(prompts)
2. Or omit metadata (pass None) if per-prompt metadata is not needed
3. Verify len(callbacks) == len(prompts) first, since the metadata check only runs in that branch

Example fix

# before
llm.generate(prompts, callbacks=per_prompt_cbs, metadata={'run': 'x'})
# after
llm.generate(prompts, callbacks=per_prompt_cbs, metadata=[{'run': 'x'}] * len(prompts))

n = len(prompts)
if metadata is not None:
    assert isinstance(metadata, list) and len(metadata) == n, f'metadata must be a list of {n} dicts'

def valid_batch_metadata(md: object, n: int) -> bool:
    return md is None or (isinstance(md, list) and len(md) == n and all(isinstance(d, dict) for d in md))

try:
    llm.generate(prompts, callbacks=cbs, metadata=metadata)
except ValueError as e:
    if 'metadata must be a list' in str(e) and isinstance(metadata, dict):
        llm.generate(prompts, callbacks=cbs, metadata=[metadata] * len(prompts))
    else:
        raise

When it happens

Trigger: Calling llm.generate(prompts, callbacks=[[cb], [cb]], metadata={'user': 'a'}) — a single dict instead of a list of dicts; or metadata=[{'user': 'a'}] with 2+ prompts. Only checked when callbacks[0] is a list/BaseCallbackManager/None.

Common situations: Copying a metadata dict used with single-prompt invoke into a batched generate call; LangSmith tracing setups that attach metadata per run and forget to fan it out per prompt.

Related errors

• tags must be a list of the same length as prompts
• run_name must be a list of the same length as prompts
• Number of manually provided run_id's does not match batch le
• Argument 'prompts' is expected to be of type list[str], rece
• invalid IP address

AI-assisted analysis of langchain-ai/langchain@e32fa9a52e (2026-08-14). Data as JSON: /api/errors/87e2cc8691a2284f. Report an issue: GitHub.