langchain-ai/langchain · error · ValueError

Number of manually provided run_id's does not match batch le

Error message

Number of manually provided run_id's does not match batch length. {len(run_id)} != {len(prompts)}

What it means

Raised by BaseLLM._get_run_ids_list when run_id is passed as a list to generate but its length differs from the number of prompts. Manually supplied run IDs must map one-to-one onto the batch so each LLM run gets a deterministic UUID.

Source

Thrown at libs/core/langchain_core/language_models/llms.py:1088

        else:
            llm_output = {}
            run_info = None
        generations = [existing_prompts[i] for i in range(len(prompts))]
        return LLMResult(generations=generations, llm_output=llm_output, run=run_info)

    @staticmethod
    def _get_run_ids_list(
        run_id: uuid.UUID | list[uuid.UUID | None] | None, prompts: list[str]
    ) -> list[uuid.UUID | None]:
        if run_id is None:
            return [None] * len(prompts)
        if isinstance(run_id, list):
            if len(run_id) != len(prompts):
                msg = (
                    "Number of manually provided run_id's does not match batch length."
                    f" {len(run_id)} != {len(prompts)}"
                )
                raise ValueError(msg)
            return run_id
        return [run_id] + [None] * (len(prompts) - 1)

    async def _agenerate_helper(
        self,
        prompts: list[str],
        stop: list[str] | None,
        run_managers: list[AsyncCallbackManagerForLLMRun],
        *,
        new_arg_supported: bool,
        **kwargs: Any,
    ) -> LLMResult:
        try:
            output = (
                await self._agenerate(
                    prompts,
                    stop=stop,
                    run_manager=run_managers[0] if run_managers else None,

View on GitHub (pinned to e32fa9a52e)

Solutions

1. Build one UUID per prompt: run_id=[uuid.uuid4() for _ in prompts]
2. Or pass a single uuid.UUID to apply it to the first prompt only
3. Or pass run_id=None to let LangChain generate IDs

Example fix

# before
llm.generate(prompts, run_id=[uuid.uuid4()])
# after
llm.generate(prompts, run_id=[uuid.uuid4() for _ in prompts])

if isinstance(run_id, list):
    assert len(run_id) == len(prompts), f'{len(run_id)} run_ids for {len(prompts)} prompts'

def valid_run_ids(run_id: object, n: int) -> bool:
    import uuid
    if run_id is None or isinstance(run_id, uuid.UUID):
        return True
    return isinstance(run_id, list) and len(run_id) == n and all(i is None or isinstance(i, uuid.UUID) for i in run_id)

try:
    llm.generate(prompts, run_id=run_ids)
except ValueError as e:
    if 'run_id' in str(e) and len(run_ids) < len(prompts):
        run_ids = run_ids + [uuid.uuid4() for _ in range(len(prompts) - len(run_ids))]
        llm.generate(prompts, run_id=run_ids)
    else:
        raise

When it happens

Trigger: llm.generate([p1, p2, p3], run_id=[uuid1, uuid2]) — 2 IDs for 3 prompts. A single UUID (not in a list) is accepted and applied to the first prompt only; only a mismatched list raises.

Common situations: Replaying or resuming batched requests with fixed run IDs (e.g. for LangSmith trace correlation or idempotent caching) and forgetting to regenerate the ID list after changing batch size.

Related errors

• tags must be a list of the same length as prompts
• metadata must be a list of the same length as prompts
• run_name must be a list of the same length as prompts
• 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/410a66ea320fb6e7. Report an issue: GitHub.