diff --git a/vllm/model_executor/layers/quantization/auto_round.py b/vllm/model_executor/layers/quantization/auto_round.py
index 0e4815be603e..f1943d461187 100644
--- a/vllm/model_executor/layers/quantization/auto_round.py
+++ b/vllm/model_executor/layers/quantization/auto_round.py
@@ -436,6 +436,12 @@ class AutoRoundConfig(QuantizationConfig):
             return None
 
     def get_quant_method(self, layer: torch.nn.Module, prefix: str):
+        if prefix and self.extra_config:
+            for layer_name in self.extra_config:
+                if (
+                    layer_name == prefix or layer_name == f"model.{prefix}"
+                ) and self.extra_config[layer_name].get("bits", 16) >= 16:
+                    return UnquantizedLinearMethod()
         if (
             current_platform.is_cpu()
             or current_platform.is_xpu()