diff --git a/litellm/router.py b/litellm/router.py
index 23618123f3..df4c2e046a 100644
--- a/litellm/router.py
+++ b/litellm/router.py
@@ -50,7 +50,6 @@ class Router:
     model_names: List = []
     cache_responses: Optional[bool] = False
     default_cache_time_seconds: int = 1 * 60 * 60  # 1 hour
-    num_retries: int = openai.DEFAULT_MAX_RETRIES
     tenacity = None
     leastbusy_logger: Optional[LeastBusyLoggingHandler] = None
     lowesttpm_logger: Optional[LowestTPMLoggingHandler] = None
@@ -237,6 +236,8 @@ class Router:
             self.num_retries = num_retries
         elif litellm.num_retries is not None:
             self.num_retries = litellm.num_retries
+        else:
+            self.num_retries = openai.DEFAULT_MAX_RETRIES
 
         self.timeout = timeout or litellm.request_timeout
 
diff --git a/litellm/tests/test_router_fallbacks.py b/litellm/tests/test_router_fallbacks.py
index 51d9451a87..364319929e 100644
--- a/litellm/tests/test_router_fallbacks.py
+++ b/litellm/tests/test_router_fallbacks.py
@@ -831,6 +831,7 @@ def test_usage_based_routing_fallbacks():
             routing_strategy="usage-based-routing",
             redis_host=os.environ["REDIS_HOST"],
             redis_port=os.environ["REDIS_PORT"],
+            num_retries=0,
         )
 
         messages = [