diff --git a/python/sglang/srt/layers/moe/ep_moe/token_dispatcher.py b/python/sglang/srt/layers/moe/ep_moe/token_dispatcher.py
index 33b7d6929..339f2e8bc 100644
--- a/python/sglang/srt/layers/moe/ep_moe/token_dispatcher.py
+++ b/python/sglang/srt/layers/moe/ep_moe/token_dispatcher.py
@@ -584,8 +584,10 @@ class _DeepEPDispatcherImplLowLatency(_DeepEPDispatcherImplBase):
                 use_fp8=use_fp8,
                 async_finish=not self.return_recv_hook,
                 return_recv_hook=self.return_recv_hook,
-                round_scale=deep_gemm_wrapper.DEEPGEMM_V202506,
-                use_ue8m0=deep_gemm_wrapper.DEEPGEMM_V202506,
+                round_scale=deep_gemm_wrapper.ENABLE_JIT_DEEPGEMM
+                and deep_gemm_wrapper.DEEPGEMM_V202506,
+                use_ue8m0=deep_gemm_wrapper.ENABLE_JIT_DEEPGEMM
+                and deep_gemm_wrapper.DEEPGEMM_V202506,
             )
         )
         return packed_recv_hidden, packed_recv_count, event, hook
diff --git a/python/sglang/srt/models/deepseek_v2.py b/python/sglang/srt/models/deepseek_v2.py
index 352480f2d..0ebc53442 100644
--- a/python/sglang/srt/models/deepseek_v2.py
+++ b/python/sglang/srt/models/deepseek_v2.py
@@ -1914,7 +1914,10 @@ class DeepseekV2ForCausalLM(nn.Module):
                 self_attn.w_vc = bind_or_assign(self_attn.w_vc, w_vc.contiguous())
                 self_attn.use_deep_gemm_bmm = True
 
-        if deep_gemm_wrapper.DEEPGEMM_SCALE_UE8M0:
+        if (
+            deep_gemm_wrapper.ENABLE_JIT_DEEPGEMM
+            and deep_gemm_wrapper.DEEPGEMM_SCALE_UE8M0
+        ):
             self._weight_requant_ue8m0()
 
     def _weight_requant_ue8m0(self):