diff --git a/litellm/proxy/db/daily_spend_bulk_upsert.py b/litellm/proxy/db/daily_spend_bulk_upsert.py
index 74d4ee29029..2c20c521fb5 100644
--- a/litellm/proxy/db/daily_spend_bulk_upsert.py
+++ b/litellm/proxy/db/daily_spend_bulk_upsert.py
@@ -215,8 +215,6 @@ def build_bulk_upsert(
         if table.carries_request_id
         else ""
     )
-    # Rows written by an older proxy, or by a request without a team's virtual key, carry "", so
-    # only a known team overwrites the stored one.
     team_id_update: Final = (
         f', "team_id" = COALESCE(NULLIF(EXCLUDED."team_id", \'\'), {quoted_table}."team_id")'
         if table.carries_team_id
diff --git a/scripts/check_schema_compatibility.py b/scripts/check_schema_compatibility.py
index 541201fede8..898c65dff2c 100644
--- a/scripts/check_schema_compatibility.py
+++ b/scripts/check_schema_compatibility.py
@@ -12,47 +12,17 @@ from typing import Final
 from pydantic import TypeAdapter
 
 Schema = Mapping[str, Mapping[str, str]]
-UniqueKeys = Mapping[str, frozenset[str]]
 MODELS: Final = re.compile(r"^[ \t]*model\s+(\w+)\s*\{(.*?)^[ \t]*\}", re.MULTILINE | re.DOTALL)
 
 
-def _without_comment(tokens: list[str]) -> list[str]:
-    return tokens[: next((i for i, token in enumerate(tokens) if token.startswith("//")), len(tokens))]
-
-
-def _is_field(part: list[str]) -> bool:
-    return len(part) >= 2 and not part[0].startswith("@@")
-
-
 def parse_model_fields(schema: str) -> Schema:
     def fields(body: str) -> Mapping[str, str]:
-        parts: Final = tuple(_without_comment(line.split()) for line in body.splitlines())
-        return MappingProxyType({part[0]: " ".join(part[1:]) for part in parts if _is_field(part)})
+        parts: Final = tuple(line.split() for line in body.splitlines())
+        return MappingProxyType({part[0]: " ".join(part[1:]) for part in parts if len(part) >= 2})
 
     return MappingProxyType({match[1]: fields(match[2]) for match in MODELS.finditer(schema)})
 
 
-def parse_unique_keys(schema: str) -> UniqueKeys:
-    def unique_keys(body: str) -> frozenset[str]:
-        parts: Final = tuple(_without_comment(line.split()) for line in body.splitlines())
-        return frozenset(" ".join(part) for part in parts if part and part[0].startswith("@@unique"))
-
-    return MappingProxyType({match[1]: unique_keys(match[2]) for match in MODELS.finditer(schema)})
-
-
-def check_unique_key_changes(old_keys: UniqueKeys, new_keys: UniqueKeys) -> tuple[str, ...]:
-    """An older proxy's upsert names its unique key in ON CONFLICT, so dropping or changing one breaks it."""
-
-    def changes() -> Iterator[str]:
-        for model, keys in old_keys.items():
-            if model not in new_keys:
-                continue
-            for key in sorted(keys - new_keys[model]):
-                yield f"Breaking: Unique key {model}.{key} was removed or changed"
-
-    return tuple(changes())
-
-
 def _field_changes(model: str, field: str, old_type: str, new_fields: Mapping[str, str]) -> Iterator[str]:
     if field not in new_fields:
         yield f"Breaking: Field {model}.{field} was removed"
@@ -93,10 +63,8 @@ def main() -> int:
     previous: Final = subprocess.run(
         ("git", "show", f"{base}:schema.prisma"), capture_output=True, text=True, check=True
     ).stdout
-    current: Final = Path("schema.prisma").read_text()
-    changes: Final = (
-        *check_breaking_changes(parse_model_fields(previous), parse_model_fields(current)),
-        *check_unique_key_changes(parse_unique_keys(previous), parse_unique_keys(current)),
+    changes: Final = check_breaking_changes(
+        parse_model_fields(previous), parse_model_fields(Path("schema.prisma").read_text())
     )
     if changes:
         sys.stderr.write("\n".join(changes) + "\n")
diff --git a/tests/unit/proxy/db/test_daily_spend_bulk_upsert.py b/tests/unit/proxy/db/test_daily_spend_bulk_upsert.py
index 1f80e184516..af2d84c5169 100644
--- a/tests/unit/proxy/db/test_daily_spend_bulk_upsert.py
+++ b/tests/unit/proxy/db/test_daily_spend_bulk_upsert.py
@@ -134,6 +134,13 @@ def test_request_id_is_preserved_when_a_later_batch_carries_none():
     assert None in params
 
 
+def test_a_batch_without_a_team_never_clears_the_stored_team():
+    sql, params = build_bulk_upsert(TAG_TABLE, merge_by_conflict_key(TAG_TABLE, (tag_txn(team_id=""),)))
+
+    assert '"team_id" = COALESCE(NULLIF(EXCLUDED."team_id", \'\'), "LiteLLM_DailyTagSpend"."team_id")' in sql
+    assert "" in params
+
+
 def test_non_tag_tables_carry_no_request_id_column():
     user_txn = {**tag_txn(), "user_id": "u-1"}
     del user_txn["tag"]
diff --git a/tests/unit/proxy/db/test_db_spend_update_writer.py b/tests/unit/proxy/db/test_db_spend_update_writer.py
index 4dfbca7db2c..29ad4979a1b 100644
--- a/tests/unit/proxy/db/test_db_spend_update_writer.py
+++ b/tests/unit/proxy/db/test_db_spend_update_writer.py
@@ -1344,6 +1344,45 @@ async def test_batch_database_updates_queues_org_member_spend_for_the_request_us
     assert transactions["org_member_list_transactions"] == {"organization_id::org1::user_id::u1": 0.1}
 
 
+@pytest.mark.asyncio
+async def test_batch_database_updates_puts_the_virtual_key_team_on_tag_rows():
+    db_writer: Final = DBSpendUpdateWriter()
+    prisma: Final = MagicMock()
+    prisma.get_request_status = MagicMock(return_value="success")
+    await db_writer._batch_database_updates(
+        response_cost=0.01,
+        user_id="u1",
+        hashed_token="a" * 64,
+        team_id="team-a",
+        org_id=None,
+        end_user_id=None,
+        prisma_client=prisma,
+        litellm_proxy_budget_name=None,
+        payload=cast(
+            SpendLogsPayload,
+            {
+                "request_id": "req-1",
+                "request_tags": '["cost-center:cc-42"]',
+                "team_id": "team-a",
+                "user": "u1",
+                "startTime": "2026-10-29T00:00:00",
+                "api_key": "a" * 64,
+                "model": "gpt-4o",
+                "custom_llm_provider": "openai",
+                "model_group": "gpt-4o",
+                "prompt_tokens": 10,
+                "completion_tokens": 5,
+                "spend": 0.01,
+                "metadata": '{"usage_object": {}}',
+            },
+        ),
+        virtual_key_team_id="team-a",
+    )
+    queued: Final = await db_writer.daily_tag_spend_update_queue.flush_and_get_aggregated_daily_spend_update_transactions()
+
+    assert [cast(DailyTagSpendTransaction, txn).get("team_id") for txn in queued.values()] == ["team-a"]
+
+
 @pytest.mark.asyncio
 async def test_project_spend_is_persisted_to_project_table_and_project_cache_is_evicted():
     db_writer: Final = DBSpendUpdateWriter()
diff --git a/tests/unit/proxy/test_db_schema_changes.py b/tests/unit/proxy/test_db_schema_changes.py
index e6f9eb20dcc..3a5d0fb1a0a 100644
--- a/tests/unit/proxy/test_db_schema_changes.py
+++ b/tests/unit/proxy/test_db_schema_changes.py
@@ -2,12 +2,7 @@ from typing import Final
 
 import pytest
 
-from scripts.check_schema_compatibility import (
-    check_breaking_changes,
-    check_unique_key_changes,
-    parse_model_fields,
-    parse_unique_keys,
-)
+from scripts.check_schema_compatibility import check_breaking_changes, parse_model_fields
 
 ORIGINAL: Final = "model Item {\n id String @id\n label String?\n}\n"
 
@@ -21,51 +16,6 @@ def test_parse_model_fields_retains_types_and_attributes() -> None:
     assert parse_model_fields(ORIGINAL) == {"Item": {"id": "String @id", "label": "String?"}}
 
 
-def test_block_attributes_and_comments_are_not_fields() -> None:
-    schema: Final = (
-        "model Item {\n id String @id\n label String?\n // free text\n"
-        ' @@unique([label, id], map: "item_label_id_key")\n @@index([label, id])\n @@index([id])\n}\n'
-    )
-    assert parse_model_fields(schema) == {"Item": {"id": "String @id", "label": "String?"}}
-
-
-@pytest.mark.parametrize(
-    "after",
-    [
-        "model Item {\n id String @id\n label String?\n @@unique([label, id, tenant])\n}\n",
-        "model Item {\n id String @id\n label String?\n}\n",
-    ],
-    ids=["changed", "removed"],
-)
-def test_changing_or_removing_a_unique_key_is_breaking(after: str) -> None:
-    before: Final = "model Item {\n id String @id\n label String?\n @@unique([label, id])\n}\n"
-    assert check_breaking_changes(parse_model_fields(before), parse_model_fields(after)) == ()
-    assert check_unique_key_changes(parse_unique_keys(before), parse_unique_keys(after)) == (
-        "Breaking: Unique key Item.@@unique([label, id]) was removed or changed",
-    )
-
-
-def test_adding_an_index_or_rewording_a_comment_is_not_breaking() -> None:
-    before: Final = (
-        'model Item {\n id String @id\n label String @default("") // old note\n'
-        " @@unique([label, id])\n @@index([id])\n}\n"
-    )
-    after: Final = (
-        'model Item {\n id String @id\n label String @default("") // reworded note\n'
-        " @@unique([label,  id]) // same key\n @@index([id])\n @@index([label, id])\n}\n"
-    )
-    assert check_breaking_changes(parse_model_fields(before), parse_model_fields(after)) == ()
-    assert check_unique_key_changes(parse_unique_keys(before), parse_unique_keys(after)) == ()
-
-
-def test_a_narrowed_column_is_still_caught_next_to_block_attributes() -> None:
-    before: Final = "model Item {\n id String @id\n label String?\n @@index([label, id])\n}\n"
-    after: Final = "model Item {\n id String @id\n label String\n @@index([label, id])\n}\n"
-    assert check_breaking_changes(parse_model_fields(before), parse_model_fields(after)) == (
-        "Breaking: Field Item.label changed from optional to required",
-    )
-
-
 def test_parse_model_fields_keeps_json_defaults_and_empty_models() -> None:
     schema: Final = 'model Item {\n data Json @default("{}")\n}\nmodel Empty {\n}\n'
     assert parse_model_fields(schema) == {"Item": {"data": 'Json @default("{}")'}, "Empty": {}}