openai/openai-python · error · TypeError

Non BaseModel types are only supported with Pydantic v2 - {m

Error message

Non BaseModel types are only supported with Pydantic v2 - {model}

What it means

to_strict_json_schema converts a type into the strict JSON schema the API needs for structured outputs. It accepts only pydantic BaseModel classes or pydantic v2 TypeAdapter instances; anything else (plain classes, pydantic v1 types, instances) raises this TypeError. Note dataclass-like types must already be wrapped in a TypeAdapter by callers such as type_to_response_format_param.

Source

Thrown at src/openai/lib/_pydantic.py:22

from typing import Any, TypeVar
from typing_extensions import TypeGuard

import pydantic

from .._types import NOT_GIVEN
from .._utils import is_dict as _is_dict, is_list
from .._compat import PYDANTIC_V1, model_json_schema

_T = TypeVar("_T")


def to_strict_json_schema(model: type[pydantic.BaseModel] | pydantic.TypeAdapter[Any]) -> dict[str, Any]:
    if inspect.isclass(model) and is_basemodel_type(model):
        schema = model_json_schema(model)
    elif (not PYDANTIC_V1) and isinstance(model, pydantic.TypeAdapter):
        schema = model.json_schema()
    else:
        raise TypeError(f"Non BaseModel types are only supported with Pydantic v2 - {model}")

    return _ensure_strict_json_schema(schema, path=(), root=schema)


def _ensure_strict_json_schema(
    json_schema: object,
    *,
    path: tuple[str, ...],
    root: dict[str, object],
) -> dict[str, Any]:
    """Mutates the given JSON schema to ensure it conforms to the `strict` standard
    that the API expects.
    """
    if not is_dict(json_schema):
        raise TypeError(f"Expected {json_schema} to be a dictionary; path={path}")

    defs = json_schema.get("$defs")
    if is_dict(defs):

View on GitHub (pinned to 9917c6e28e)

Solutions

1. Pass the pydantic BaseModel class itself (not an instance)
2. For dataclass/TypedDict types, wrap with pydantic.TypeAdapter (requires pydantic v2)
3. Use the SDK helpers pydantic_function_tool / type_to_response_format_param which do the wrapping for you

Example fix

# before
schema = to_strict_json_schema(MyDataclass)
# after
schema = to_strict_json_schema(pydantic.TypeAdapter(MyDataclass))

def strict_schema_input_ok(obj: object) -> bool:
    if inspect.isclass(obj):
        return is_basemodel_type(obj)
    return (not PYDANTIC_V1) and isinstance(obj, pydantic.TypeAdapter)

When it happens

Trigger: Calling to_strict_json_schema(SomePlainClass), passing a BaseModel instance instead of the class, or passing a TypeAdapter under pydantic v1 (where the isinstance check is short-circuited by 'not PYDANTIC_V1').

Common situations: Direct use of the helper with unsupported types; pydantic v1 environments; passing a class that inherits from something BaseModel-like but is not a real BaseModel.

Related errors

• Expected {json_schema} to be a dictionary; path={path}
• Expected `$ref: {ref}` to resolved to a dictionary but got {
• Unexpected $ref format {ref!r}; Does not start with #/
• Pydantic models must subclass our base model type, e.g. `fro
• mode must be either 'json' or 'python'

AI-assisted analysis of openai/openai-python@9917c6e28e (2026-08-28). Data as JSON: /api/errors/11ec3ef1a26a787b. Report an issue: GitHub.