diff --git a/python/sglang/srt/layers/quantization/moe_wna16.py b/python/sglang/srt/layers/quantization/moe_wna16.py
index 531e4271f..ce6397dd1 100644
--- a/python/sglang/srt/layers/quantization/moe_wna16.py
+++ b/python/sglang/srt/layers/quantization/moe_wna16.py
@@ -18,7 +18,10 @@ from sglang.srt.layers.quantization.base_config import (
     QuantizeMethodBase,
 )
 from sglang.srt.layers.quantization.gptq import GPTQConfig, GPTQMarlinConfig
-from sglang.srt.layers.quantization.unquant import UnquantizedLinearMethod
+from sglang.srt.layers.quantization.unquant import (
+    UnquantizedFusedMoEMethod,
+    UnquantizedLinearMethod,
+)
 from sglang.srt.utils import get_device_capability, set_weight_attrs
 
 logger = logging.getLogger(__name__)
@@ -194,6 +197,8 @@ class MoeWNA16Config(QuantizationConfig):
         from sglang.srt.layers.moe.fused_moe_triton.layer import FusedMoE
 
         if is_layer_skipped_quant(prefix, self.modules_to_not_convert):
+            if isinstance(layer, FusedMoE):
+                return UnquantizedFusedMoEMethod()
             return UnquantizedLinearMethod()
         elif isinstance(layer, LinearBase):