pydantic_compatibilityTier 1 · 70% confidence

ai-agents-pydantic-compatibili-using-pydanticoutputparser-or-create-structured-ou-9aa5f6f6

agent: ai_agents

When does this happen?

IF Using PydanticOutputParser (or create_structured_output_runnable) with a pydantic v2 BaseModel in langchain >=0.1.6 causes ValidationError: 'subclass of BaseModel expected'.

How others solved it

THEN Downgrade langchain to 0.1.5, or ensure your output model inherits from pydantic.v1.BaseModel (via langchain's pydantic v1 compatibility layer) instead of pydantic v2 BaseModel.

from langchain.output_parsers import PydanticOutputParser
from pydantic import BaseModel

class TestModel(BaseModel):
    test_attribute: str

# This fails in langchain >=0.1.6
parser = PydanticOutputParser(pydantic_object=TestModel)

# Workaround: use pydantic v1 BaseModel
# from pydantic.v1 import BaseModel as BaseModelV1
# class TestModel(BaseModelV1):
#     test_attribute: str

Related patterns

Have you seen this in your site?

Connect AgentMinds to match against your tech stack automatically.

Run diagnostics