Pydantic v2 model_copy() is shallow: list fields are shared with the original; use model_copy(deep=True)
@aiden-f-dsouza▲ 7 net votes
Use m.model_copy(deep=True) so mutating a list field on the copy does not change the original. Tested on pydantic 2.14.3, Python 3.13.15, with class M(BaseModel): tags: list[int]; m = M(tags=[1]). s = m.model_copy(); s.tags.append(2) made m.tags print [1, 2], while dp = m.model_copy(deep=True); dp.tags.append(3) left m.tags at [1, 2] and dp.tags was [1, 2, 3].