fix(proxy): keep the token's team model list in the websocket passthrough gate without a database

litellm/proxy/auth/auth_checks.py

@@ -4164,7 +4164,7 @@ async def enforced_model_allowlists(
     """One model allowlist per level that ``common_checks`` enforces on a request from this identity."""
     key_models: Final = _resolve_key_models_for_auth_check(valid_token=valid_token)
     if prisma_client is None:
-        return (key_models,)
+        return (key_models, tuple(valid_token.team_models or ()))
     team_object: Final = (
         None
         if valid_token.team_id is None

tests/test_litellm/proxy/auth/test_auth_checks.py

@@ -7392,7 +7392,13 @@ async def test_enforced_model_allowlists_reads_every_level_from_cache():
         proxy_logging_obj=proxy_logging_obj,
     )
     without_database = await enforced_model_allowlists(
-        valid_token=UserAPIKeyAuth(token="hashed-fake", models=["gpt-4o"], user_id="user-fake", team_id="team-fake"),
+        valid_token=UserAPIKeyAuth(
+            token="hashed-fake",
+            models=["gpt-4o"],
+            team_models=["gpt-4o-mini"],
+            user_id="user-fake",
+            team_id="team-fake",
+        ),
         prisma_client=None,
         user_api_key_cache=cache,
         proxy_logging_obj=proxy_logging_obj,
@@ -7406,4 +7412,4 @@ async def test_enforced_model_allowlists_reads_every_level_from_cache():
         ["gpt-4.1"],
     ]
     assert [list(scope) for scope in personal] == [[], [], [], ["o3"], []]
-    assert [list(scope) for scope in without_database] == [["gpt-4o"]]
+    assert [list(scope) for scope in without_database] == [["gpt-4o"], ["gpt-4o-mini"]]

tests/test_litellm/proxy/test_openai_ws_passthrough_routes.py

@@ -14,6 +14,7 @@ from litellm.proxy._types import UserAPIKeyAuth
 from litellm.proxy.pass_through_endpoints.llm_passthrough_endpoints import (
     _OPENAI_WS_DISABLED_REFUSAL,
     _OPENAI_WS_MODEL_RESTRICTED_REFUSAL,
+    _has_model_restrictions,
     _openai_websocket_refusal,
     _proxy_model_allowlists,
     openai_websocket_proxy_route,
@@ -288,8 +289,10 @@ async def test_openai_websocket_allows_unrestricted_identities(scopes):
 
 
 @pytest.mark.asyncio
-async def test_proxy_model_allowlists_reads_the_key_scope_without_a_database():
+async def test_proxy_model_allowlists_reads_the_token_scopes_without_a_database():
+    token: Final = UserAPIKeyAuth(models=[], team_id="team-fake", team_models=["gpt-4o"])
     with patch("litellm.proxy.proxy_server.prisma_client", None):
-        scopes = await _proxy_model_allowlists()(UserAPIKeyAuth(models=["gpt-4o"]))
+        scopes = await _proxy_model_allowlists()(token)
 
-    assert tuple(tuple(scope) for scope in scopes) == (("gpt-4o",),)
+    assert tuple(tuple(scope) for scope in scopes) == ((), ("gpt-4o",))
+    assert _has_model_restrictions(scopes)