SQLModel fails to create DB constraints when using Annotated types after Pydantic 2.12.0

[YuriiMotov]

Discussed in #1597

^{Originally posted by MarishLakshmanan October 8, 2025}

First Check

• I added a very descriptive title here.
• I used the GitHub search to find a similar question and didn't find it.
• I searched the SQLModel documentation, with the integrated search.
• I already searched in Google "How to X in SQLModel" and didn't find any information.
• I already read and followed all the tutorial in the docs and didn't find an answer.
• I already checked if it is not related to SQLModel but to Pydantic.
• I already checked if it is not related to SQLModel but to SQLAlchemy.

Commit to Help

• I commit to help with one of those options 👆

Example Code

Environment:
Pydantic==2.12.0

Code:

from pydantic import StringConstraints
NonEmptyString = Annotated[str, StringConstraints(strip_whitespace=True, min_length=1)]

Class Test(SQLModel):
 id: NonEmptyString = Field(primary_key=True) # this was working fine on pydantic v2.11.9

Description

Description

After upgrading to Pydantic 2.12.0, SQLModel no longer creates database constraints (like primary keys, unqiue keys ) for fields defined using Annotated types.

Example:

from typing import Annotated
from sqlmodel import SQLModel, Field
from pydantic import StringConstraints

NonEmptyString = Annotated[str, StringConstraints()]

class Test(SQLModel, table=True):
 id: NonEmptyString = Field(primary_key=True)

Expected Behavior

The table should be created with id as an varchar and primary key

Actual Behavior

it throws an error

sqlalchemy.exc.ArgumentError: Mapper Mapper[Test(test)] could not assemble any primary key columns for mapped table 'test'

Additional notes

In case you use an unique key constraint the table will be created, you will not get any error but the unique key constraint will not be created.

Operating System

Linux

Operating System Details

No response

SQLModel Version

0.0.26

Python Version

3.13.3

Additional Context

No response

[YuriiMotov]

Test taken from #1603

from sqlmodel import Field, SQLModel
from typing_extensions import Annotated


def test_declaration_syntax_1():
 class Person1(SQLModel):
 name: str = Field(primary_key=True)

 class Person1Final(Person1, table=True):
 pass


def test_declaration_syntax_2():
 class Person2(SQLModel):
 name: Annotated[str, Field(primary_key=True)]

 class Person2Final(Person2, table=True):
 pass


def test_declaration_syntax_3():
 class Person3(SQLModel):
 name: Annotated[str, ...] = Field(primary_key=True)

 class Person3Final(Person3, table=True):
 pass

test_declaration_syntax_2 and test_declaration_syntax_3 work well with Pydantic 2.11.10, but fail with Pydantic 2.12

PRs:

• 🐛 Fix support for Annotated fields with Pydantic 2.12+ #1603
• 🐛 Fix support for Annotated fields with Pydantic 2.12+ #1607

[Samreay]

I think I just ran into this issue too, but in relation to timestamps with timezone:

from datetime import datetime as dt
from datetime import timezone as tz
from typing_extensions import Annotated

from sqlmodel import TIMESTAMP, Field, SQLModel
from pydantic import BeforeValidator

TIMESTAMP_WITH_TZ: type = TIMESTAMP(timezone=True)

IntDatetime = Annotated[
 dt,
 BeforeValidator(lambda x: x if isinstance(x, dt) else dt.fromtimestamp(int(x) / 1000, tz=tz.utc)),
]

# This model works with both 2.11.10 and 2.12.0
class Message(SQLModel, table=True):
 time: dt = Field(sa_type=TIMESTAMP_WITH_TZ)

# This model works with <=2.11.10, but 2.12.0 generates an error
# `can't subtract offset-naive and offset-aware datetimes`
# As the SQL generated is $1::TIMESTAMP WITHOUT TIME ZONE
class Message(SQLModel, table=True):
 time: IntDatetime = Field(sa_type=TIMESTAMP_WITH_TZ)

[KrilleGH]

I wanted to share this example which can also be used as a test.
With Pydantic 2.12 it prints an empty set while it should print:
{ForeignKey('vehicle.id')}

from pydantic import PositiveInt

from sqlmodel import Field, Session, SQLModel, col, select


class Vehicle(SQLModel, table=True):
 name: str
 id: PositiveInt | None = Field(default=None, primary_key=True)


class Drivers_Logbook(SQLModel, table=True):
 # To make this example succeed, replace the next line with either
 # vehicle_id: int = Field(foreign_key='vehicle.id')
 # vehicle_id: PositiveInt | None = Field(foreign_key='vehicle.id', default=None)
 vehicle_id: PositiveInt = Field(foreign_key='vehicle.id')
 id: PositiveInt | None = Field(default=None, primary_key=True)


drivers_logbook = SQLModel.metadata.tables['drivers_logbook']
print(drivers_logbook.foreign_keys)

I also have a version using pure SQLAlchemy and couldn't reproduce the issue there.

[dwreeves]

Ran into this. The fundamental issue appears to be that the subclass sqlmodel.main.FieldInfo is no longer respected in Pydantic 2.12+. The object being passed to get_column_from_field is of type pydantic.fields.FieldInfo, whereas in previous versions it was the sqlmodel subclass. #1603 seems like a reasonable resolution as it restores the original sqlmodel.main.FieldInfo object to pass into everything else.

[frgfm]

I just found out I had the same problem: no issue with pydantic==2.11.9, but trouble after that. For reference, in both cases I have the same version of sqlalchemy==2.0.44 and sqlmodel==0.0.27 in a locked container.

My snippet for the regression:

import uuid
from sqlmodel import Field, SQLModel

class User(SQLModel, table=True):
 __tablename__ = "users"
 id: Annotated[UUID4 | None, Field(default_factory=uuid.uuid4, primary_key=True)]
 name: str

the error with pydantic==2.12.4:

Traceback (most recent call last):
 File "/app/test.py", line 4, in <module>
 class User(SQLModel, table=True):
 File "/opt/venv/lib/python3.11/site-packages/sqlmodel/main.py", line 644, in __init__
 DeclarativeMeta.__init__(cls, classname, bases, dict_, **kw)
 File "/opt/venv/lib/python3.11/site-packages/sqlalchemy/orm/decl_api.py", line 199, in __init__
 _as_declarative(reg, cls, dict_)
 File "/opt/venv/lib/python3.11/site-packages/sqlalchemy/orm/decl_base.py", line 245, in _as_declarative
 return _MapperConfig.setup_mapping(registry, cls, dict_, None, {})
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 File "/opt/venv/lib/python3.11/site-packages/sqlalchemy/orm/decl_base.py", line 326, in setup_mapping
 return _ClassScanMapperConfig(
 ^^^^^^^^^^^^^^^^^^^^^^^
 File "/opt/venv/lib/python3.11/site-packages/sqlalchemy/orm/decl_base.py", line 581, in __init__
 self._early_mapping(mapper_kw)
 File "/opt/venv/lib/python3.11/site-packages/sqlalchemy/orm/decl_base.py", line 367, in _early_mapping
 self.map(mapper_kw)
 File "/opt/venv/lib/python3.11/site-packages/sqlalchemy/orm/decl_base.py", line 1995, in map
 mapper_cls(self.cls, self.local_table, **self.mapper_args),
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 File "<string>", line 2, in __init__
 File "/opt/venv/lib/python3.11/site-packages/sqlalchemy/util/deprecations.py", line 281, in warned
 return fn(*args, **kwargs) # type: ignore[no-any-return]
 ^^^^^^^^^^^^^^^^^^^
 File "/opt/venv/lib/python3.11/site-packages/sqlalchemy/orm/mapper.py", line 866, in __init__
 self._configure_pks()
 File "/opt/venv/lib/python3.11/site-packages/sqlalchemy/orm/mapper.py", line 1652, in _configure_pks
 raise sa_exc.ArgumentError(
sqlalchemy.exc.ArgumentError: Mapper Mapper[User(users)] could not assemble any primary key columns for mapped table 'users'