diff --git a/vllm/config/model.py b/vllm/config/model.py
index 7d2409d70..b12202f9c 100644
--- a/vllm/config/model.py
+++ b/vllm/config/model.py
@@ -2021,6 +2021,15 @@ def _get_and_verify_max_len(
 
                 if rope_type == "yarn":
                     derived_max_model_len = rp["original_max_position_embeddings"]
+        if scaling_factor is None:
+            # Fallback the factor to 1.0 if a user assigned `null`
+            logger.warning_once(
+                "The model's RoPE configuration has a null scaling "
+                "factor which is unexpected. This likely indicates a bug "
+                "in the model's HuggingFace config.json. Please notify the "
+                "model vendor. Falling back the value to 1.0. "
+            )
+            scaling_factor = 1.0
         # Do this outside loop since all layer types should have the same scaling
         derived_max_model_len *= scaling_factor