diff --git a/python/sglang/srt/layers/moe/fused_moe_triton/layer.py b/python/sglang/srt/layers/moe/fused_moe_triton/layer.py
index 80735f3da..f04479dbb 100644
--- a/python/sglang/srt/layers/moe/fused_moe_triton/layer.py
+++ b/python/sglang/srt/layers/moe/fused_moe_triton/layer.py
@@ -1376,18 +1376,18 @@ class FlashInferFP4MoE(FusedMoE):
             hidden_states_scale=hs_scale_linear.view(torch.float8_e4m3fn).reshape(
                 *hs_scale_linear.shape[:-1], -1
             ),
-            gemm1_weights=self.gemm1_weights_fp4_shuffled.data,
-            gemm1_weights_scale=self.gemm1_scales_fp4_shuffled.data.view(
-                torch.float8_e4m3fn
-            ),
+            # B300 port: HEAD's align_fp4_moe_weights_for_flashinfer_trtllm now
+            # stores the prepared (shuffled) fp4 weights/scales into the standard
+            # param names instead of gemm*_fp4_shuffled (weights=uint8, scales
+            # already fp8_e4m3fn — the .view below is a no-op kept for clarity).
+            gemm1_weights=self.w13_weight.data,
+            gemm1_weights_scale=self.w13_weight_scale.data.view(torch.float8_e4m3fn),
             gemm1_bias=None,
             gemm1_alpha=None,
             gemm1_beta=None,
             gemm1_clamp_limit=None,
-            gemm2_weights=self.gemm2_weights_fp4_shuffled.data,
-            gemm2_weights_scale=self.gemm2_scales_fp4_shuffled.data.view(
-                torch.float8_e4m3fn
-            ),
+            gemm2_weights=self.w2_weight.data,
+            gemm2_weights_scale=self.w2_weight_scale.data.view(torch.float8_e4m3fn),
             gemm2_bias=None,
             output1_scale_scalar=self.g1_scale_c.data,
             output1_scale_gate_scalar=self.g1_alphas.data,