diff --git a/vllm/model_executor/layers/fused_moe/activation.py b/vllm/model_executor/layers/fused_moe/activation.py
index 46a9bd1b4c58..bcee30e6e5c7 100644
--- a/vllm/model_executor/layers/fused_moe/activation.py
+++ b/vllm/model_executor/layers/fused_moe/activation.py
@@ -272,7 +272,7 @@ def _apply_moe_activation_masked(
             "SWIGLUOAI_UNINTERLEAVE requires clamp_limit"
         )
     elif activation == MoEActivation.SWIGLUSTEP:
-        masked_clamp_limit = 7.0
+        masked_clamp_limit = 7.0 if config.clamp_limit is None else config.clamp_limit
 
     torch.ops._C.masked_moe_activation(
         output,
@@ -390,7 +390,8 @@ def apply_moe_activation(
     elif activation == MoEActivation.SWIGLUSTEP:
         from vllm.model_executor.layers.activation import swiglustep_and_mul_triton
 
-        swiglustep_and_mul_triton(output, input)
+        limit = 7.0 if config.clamp_limit is None else config.clamp_limit
+        swiglustep_and_mul_triton(output, input, limit=limit)
 
     # Activations without gated multiplication
     elif activation == MoEActivation.SILU_NO_MUL:
diff --git a/vllm/model_executor/models/bailing_moe_v3.py b/vllm/model_executor/models/bailing_moe_v3.py
index cbb1e8bc8f3f..1e55bc8b7e57 100644
--- a/vllm/model_executor/models/bailing_moe_v3.py
+++ b/vllm/model_executor/models/bailing_moe_v3.py
@@ -1203,7 +1203,9 @@ def __init__(
             intermediate_size=config.moe_intermediate_size,
             renormalize=True,
             quant_config=quant_config,
-            activation="silu",
+            activation=(
+                "swiglustep" if self.expert_swiglu_limit is not None else "silu"
+            ),
             swiglu_limit=self.expert_swiglu_limit,
             prefix=f"{prefix}.experts",
             scoring_func="sigmoid",