diff --git a/vllm/models/kimi_k3/nvidia/model.py b/vllm/models/kimi_k3/nvidia/model.py
index a970479da8..737184b9d8 100644
--- a/vllm/models/kimi_k3/nvidia/model.py
+++ b/vllm/models/kimi_k3/nvidia/model.py
@@ -677,7 +677,7 @@ class KimiMoE(nn.Module):
                 hidden_size,
                 self.moe_hidden_size,
                 bias=False,
-                quant_config=None,
+                quant_config=quant_config,
                 prefix=f"{prefix}.routed_expert_down_proj",
             )
             self.routed_expert_norm = (
@@ -694,7 +694,7 @@ class KimiMoE(nn.Module):
                 self.moe_hidden_size,
                 hidden_size,
                 bias=False,
-                quant_config=None,
+                quant_config=quant_config,
                 prefix=f"{prefix}.routed_expert_up_proj",
             )