diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json
index 6804d677e6a..2831f1a5cc9 100644
--- a/litellm/model_prices_and_context_window_backup.json
+++ b/litellm/model_prices_and_context_window_backup.json
@@ -2454,6 +2454,17 @@
         "mode": "chat",
         "source": "https://cloud.google.com/vertex-ai/generative-ai/pricing#partner-models"
     },
+    "vertex_ai/meta/llama-3.2-90b-vision-instruct-maas": {
+        "max_tokens": 8192,
+        "max_input_tokens": 128000,
+        "max_output_tokens": 8192,
+        "input_cost_per_token": 0.0,
+        "output_cost_per_token": 0.0,
+        "litellm_provider": "vertex_ai-llama_models",
+        "mode": "chat",
+        "supports_system_messages": true,
+        "source": "https://console.cloud.google.com/vertex-ai/publishers/meta/model-garden/llama-3.2-90b-vision-instruct-maas"
+    },
     "vertex_ai/mistral-large@latest": {
         "max_tokens": 8191,
         "max_input_tokens": 128000,
diff --git a/litellm/proxy/hooks/parallel_request_limiter.py b/litellm/proxy/hooks/parallel_request_limiter.py
index f5c45f3bd02..7b3c0960ff0 100644
--- a/litellm/proxy/hooks/parallel_request_limiter.py
+++ b/litellm/proxy/hooks/parallel_request_limiter.py
@@ -322,8 +322,9 @@ class _PROXY_MaxParallelRequestsHandler(CustomLogger):
         # check if REQUEST ALLOWED for user_id
         user_id = user_api_key_dict.user_id
         if user_id is not None:
+            request_count_api_key = f"{user_id}::{precise_minute}::request_count"
             _user_id_rate_limits = await self.internal_usage_cache.async_get_cache(
-                key=user_id,
+                key=request_count_api_key,
                 litellm_parent_otel_span=user_api_key_dict.parent_otel_span,
             )
             # get user tpm/rpm limits