diff --git a/python/sglang/srt/layers/moe/moe_runner/aiter.py b/python/sglang/srt/layers/moe/moe_runner/aiter.py
index ed402a283ed7..9bc7486eba65 100644
--- a/python/sglang/srt/layers/moe/moe_runner/aiter.py
+++ b/python/sglang/srt/layers/moe/moe_runner/aiter.py
@@ -57,6 +57,7 @@ class AiterMoeQuantInfo(MoeQuantInfo):
     hidden_pad: int = 0
     intermediate_pad: int = 0
     swiglu_limit: float = 0.0
+    gate_mode: str = "separated"
 
 
 @dataclass
@@ -130,8 +131,13 @@ def run(
             extra["num_local_tokens"] = runner_input.num_local_tokens
         if runner_input.output_dtype is not None:
             extra["dtype"] = runner_input.output_dtype
+        # Single source of truth for gate_mode. The MXFP4 (GPT-OSS) path sets
+        # quant_info.gate_mode from the activation type; the FP8/DSv4 path drives
+        # it via swiglu_limit. swiglu_limit > 0 implies the interleaved gate/up
+        # layout, so force INTERLEAVE there to preserve the original behavior.
+        gate_mode = quant_info.gate_mode
         if quant_info.swiglu_limit > 0:
-            extra["gate_mode"] = GateMode.INTERLEAVE.value
+            gate_mode = GateMode.INTERLEAVE.value
             extra["swiglu_limit"] = quant_info.swiglu_limit
 
         output = fused_moe(
@@ -152,6 +158,7 @@ def run(
             doweight_stage1=quant_info.doweight_stage1,
             hidden_pad=quant_info.hidden_pad,
             intermediate_pad=quant_info.intermediate_pad,
+            gate_mode=gate_mode,
             **extra,
         )
         return AiterRunnerOutput(hidden_states=output)
diff --git a/python/sglang/srt/layers/quantization/mxfp4.py b/python/sglang/srt/layers/quantization/mxfp4.py
index 6ac793d0270f..0fed7c9005e0 100644
--- a/python/sglang/srt/layers/quantization/mxfp4.py
+++ b/python/sglang/srt/layers/quantization/mxfp4.py
@@ -1221,6 +1221,8 @@ def apply(
             )[0]
             return StandardCombineInput(hidden_states=trtllm_gen_output)
         if _use_aiter:
+            from aiter.ops.flydsl.moe_common import GateMode
+
             from sglang.srt.layers.moe.moe_runner.aiter import (
                 AiterMoeQuantInfo,
                 AiterQuantType,
@@ -1248,6 +1250,11 @@ def apply(
                 doweight_stage1=self.moe_runner_config.apply_router_weight_on_input,
                 hidden_pad=self.hidden_pad,
                 intermediate_pad=self.intermediate_pad,
+                gate_mode=(
+                    GateMode.INTERLEAVE.value
+                    if self.runner.config.activation == "swiglu"
+                    else GateMode.SEPARATED.value
+                ),
             )
             return self.runner.run(
                 dispatch_output._replace(hidden_states=x_padded), quant_info