langchain-ai/langchain · error · ValueError

Unsupported Pydantic schema: {pydantic_schema}

Error message

Unsupported Pydantic schema: {pydantic_schema}

What it means

Defensive branch on the multi-schema path of PydanticOutputFunctionsParser.parse_result: after selecting the schema (from the dict by function name, or the single schema), it is neither a pydantic v2 BaseModel nor a v1 BaseModelV1 subclass, so its arguments cannot be validated. Like error 187, this is marked unreachable for well-typed inputs.

Source

Thrown at libs/core/langchain_core/output_parsers/openai_functions.py:302

                msg = (  # type: ignore[unreachable]
                    "Unsupported Pydantic schema with args_only: "
                    f"{self.pydantic_schema}"
                )
                raise ValueError(msg)
        else:
            fn_name = result_["name"]
            args = result_["arguments"]
            if isinstance(self.pydantic_schema, dict):
                pydantic_schema = self.pydantic_schema[fn_name]
            else:
                pydantic_schema = self.pydantic_schema
            if issubclass(pydantic_schema, BaseModel):
                pydantic_args = pydantic_schema.model_validate_json(args)
            elif issubclass(pydantic_schema, BaseModelV1):
                pydantic_args = pydantic_schema.parse_raw(args)
            else:
                msg = f"Unsupported Pydantic schema: {pydantic_schema}"  # type: ignore[unreachable]
                raise ValueError(msg)
        return pydantic_args


class PydanticAttrOutputFunctionsParser(PydanticOutputFunctionsParser):
    """Parse an output as an attribute of a Pydantic object."""

    attr_name: str
    """The name of the attribute to return."""

    @override
    def parse_result(self, result: list[Generation], *, partial: bool = False) -> Any:
        """Parse the result of an LLM call to a JSON object.

        Args:
            result: The result of the LLM call.
            partial: Whether to parse partial JSON objects.

        Returns:

View on GitHub (pinned to e32fa9a52e)

Solutions

1. Ensure every value in the pydantic_schema dict is a Pydantic BaseModel subclass (v1 or v2)
2. Add an init-time check over dict values (all(issubclass(v, BaseModel) for v in schema.values())) to fail fast instead of at parse time

Example fix

# before
pydantic_schema = {"create_person": PersonModel, "create_book": BookDataclass}

# after
from pydantic import BaseModel
class Book(BaseModel):
    title: str
pydantic_schema = {"create_person": PersonModel, "create_book": Book}

from pydantic import BaseModel, v1
ok = all(
    isinstance(v, type) and (issubclass(v, BaseModel) or issubclass(v, v1.BaseModel))
    for v in pydantic_schema.values()
)

from pydantic import BaseModel, v1

def all_valid_schemas(d: dict) -> bool:
    return all(isinstance(v, type) and (issubclass(v, BaseModel) or issubclass(v, v1.BaseModel)) for v in d.values())

When it happens

Trigger: A dict schema mapping function names to non-BaseModel classes (dataclasses, plain classes), or values replaced at runtime after validator checks.

Common situations: Hand-built schema dicts mixing BaseModel subclasses with helper classes; plugin systems registering arbitrary callables as 'schemas'.

Related errors

• Unsupported Pydantic schema with args_only: {self.pydantic_s
• If multiple pydantic schemas are provided then args_only sho
• Dict Pydantic schema unsupported with args_only: {self.pydan
• Unsupported model version for PydanticOutputParser: {self.py
• Either data or path must be provided

AI-assisted analysis of langchain-ai/langchain@e32fa9a52e (2026-08-14). Data as JSON: /api/errors/bf3457bee1bf1d59. Report an issue: GitHub.