Pydantic from __future__ import annotations no longer works

The Fix

pip install pydantic==2.9.2

Based on closed pydantic/pydantic issue #10357 · PR/commit linked

Jump to Verify Open PR/Commit

@@ -448,11 +448,15 @@ def inspect_namespace(  # noqa C901
                 frame = sys._getframe(2)
                 if frame is not None:
-                    ann_type = eval_type_backport(
-                        _make_forward_ref(ann_type, is_argument=False, is_class=True),
-                        globalns=frame.f_globals,

repro.py

from __future__ import annotations

from pydantic import BaseModel


class ModelB(BaseModel):
    _foo: ModelA

class ModelA(BaseModel):
    bar: int


def main():
    model_a=ModelA(bar=1)
    model_b=ModelB(_foo=model_a)

if __name__ == "__main__":
    main()

verify

Re-run the minimal reproduction on your broken version, then apply the fix and re-run.

fix.md

Option A — Upgrade to fixed release\npip install pydantic==2.9.2\nWhen NOT to use: Do not use this fix if your code relies on the previous behavior of private attributes.\n\n

Why This Fix Works in Production

Trigger: class ModelB(BaseModel):
Mechanism: Fixes an issue where using 'from __future__ import annotations' with private attributes caused a NameError in Pydantic v2.9.0.
Why the fix works: Fixes an issue where using 'from __future__ import annotations' with private attributes caused a NameError in Pydantic v2.9.0. (first fixed release: 2.9.2).

Production impact:

If left unfixed, this can cause silent data inconsistencies that propagate (bad cache entries, incorrect downstream decisions).

Why This Breaks in Prod

Shows up under Python 3.12 in real deployments (not just unit tests).
Surfaces as: Traceback (most recent call last):

Proof / Evidence

GitHub issue: #10357
Fix PR: https://github.com/pydantic/pydantic/pull/10358
First fixed release: 2.9.2
Reproduced locally: No (not executed)
Last verified: 2026-02-09
Confidence: 0.95
Did this fix it?: Yes (upstream fix exists)
Own content ratio: 0.33

Discussion

High-signal excerpts from the issue thread (symptoms, repros, edge-cases).

Jump to Sources Open on GitHub

“### Initial Checks - [X] I confirm that I'm using Pydantic V2 ### Description Hi all! Thanks for this amazing package, it has really save me SOOO much trouble in recent times. I've just updates from v2.8.2 to v2.9.0 and I've noticed that fr”

Issue thread · issue description · source

Failure Signature (Search String)

class ModelB(BaseModel):

Error Message

Stack trace

error.txt

Error Message
-------------
Traceback (most recent call last):
  File "/home/joao/pydantic_test.py", line 6, in <module>
    class ModelB(BaseModel):
  File "/home/joao/.venv/lib/python3.12/site-packages/pydantic/_internal/_model_construction.py", line 114, in __new__
    private_attributes = inspect_namespace(
                         ^^^^^^^^^^^^^^^^^^
  File "/home/joao/.venv/lib/python3.12/site-packages/pydantic/_internal/_model_construction.py", line 448, in inspect_namespace
    ann_type = eval_type_backport(
               ^^^^^^^^^^^^^^^^^^^
  File "/home/joao/.venv/lib/python3.12/site-packages/pydantic/_internal/_typing_extra.py", line 279, in eval_type_backport
    return _eval_type_backport(value, globalns, localns, type_params)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/joao/.venv/lib/python3.12/site-packages/pydantic/_internal/_typing_extra.py", line 303, in _eval_type_backport
    return _eval_type(value, globalns, localns, type_params)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/joao/.venv/lib/python3.12/site-packages/pydantic/_internal/_typing_extra.py", line 332, in _eval_type
    return typing._eval_type(  # type: ignore
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/typing.py", line 415, in _eval_type
    return t._evaluate(globalns, localns, type_params, recursive_guard=recursive_guard)
.
... (truncated) ...

Minimal Reproduction

repro.py

from __future__ import annotations

from pydantic import BaseModel


class ModelB(BaseModel):
    _foo: ModelA

class ModelA(BaseModel):
    bar: int


def main():
    model_a=ModelA(bar=1)
    model_b=ModelB(_foo=model_a)

if __name__ == "__main__":
    main()

Environment

Python: 3.12
Pydantic: 2

What Broke

Code raises NameError when using private attributes in models, causing application crashes.

Fix Options (Details)

Option A — Upgrade to fixed release Safe default (recommended)

pip install pydantic==2.9.2

When NOT to use: Do not use this fix if your code relies on the previous behavior of private attributes.

Use when you can deploy the upstream fix. It is usually lower-risk than long-lived workarounds.

Fix reference: https://github.com/pydantic/pydantic/pull/10358

First fixed release: 2.9.2

Last verified: 2026-02-09. Validate in your environment.

When NOT to Use This Fix

Do not use this fix if your code relies on the previous behavior of private attributes.

Verify Fix

verify

Re-run the minimal reproduction on your broken version, then apply the fix and re-run.

Did This Fix Work in Your Case?

Quick signal helps us prioritize which fixes to verify and improve.

Prevention

Add a CI check that diffs key outputs after upgrades (OpenAPI schema snapshots, JSON payload shapes, CLI output).
Upgrade behind a canary and run integration tests against the canary before 100% rollout.

Version Compatibility Table

Version	Status
2.9.2	Fixed

Related Issues

No related fixes found.

Cluster: pydantic:data-consistency Pydantic hub Pydantic best practices All hubs All clusters

Related clusters: Configuration error OpenAPI schema Serialization

Sources

We don’t republish the full GitHub discussion text. Use the links above for context.

Pydantic from __future__ import annotations no longer works on v2.9.0 (Fix)