diff --git a/vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe.py b/vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe.py
index 097d0bc01891..77aedcc471fc 100644
--- a/vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe.py
+++ b/vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors_moe.py
@@ -937,7 +937,7 @@ def process_weights_after_loading(self, layer: FusedMoE) -> None:
                 w13,
                 w13_scale,
                 shard_size=layer.intermediate_size_per_partition,
-                num_experts=layer.num_local_experts,
+                num_experts=layer.local_num_experts,
                 is_act_and_mul=self.moe.is_act_and_mul,
             )