diff --git a/tests/integration/test_lists/test-db/l0_dgx_b200.yml b/tests/integration/test_lists/test-db/l0_dgx_b200.yml
index b48e1b4ee8d..4dc08bf942f 100644
--- a/tests/integration/test_lists/test-db/l0_dgx_b200.yml
+++ b/tests/integration/test_lists/test-db/l0_dgx_b200.yml
@@ -15,6 +15,7 @@ l0_dgx_b200:
       backend: pytorch
   tests:
   - unittest/_torch/multi_gpu_modeling -k "deepseek"
+  - unittest/_torch/modules/test_fused_moe.py::test_fused_moe_alltoall_fp4[MNNVL]
   - unittest/_torch/modules/test_fused_moe.py::test_fused_moe_alltoall_fp4[DeepEPLowLatency]
   - accuracy/test_llm_api_pytorch.py::TestLlama3_1_8BInstruct::test_bfloat16_4gpus[pp4-attn_backend=TRTLLM-torch_compile=False]
   - accuracy/test_llm_api_pytorch.py::TestLlama3_1_8BInstruct::test_fp8_4gpus[tp4-fp8kv=True-attn_backend=TRTLLM-torch_compile=False]
diff --git a/tests/integration/test_lists/test-db/l0_dgx_h100.yml b/tests/integration/test_lists/test-db/l0_dgx_h100.yml
index 38ea6fe8f69..92f4052f897 100644
--- a/tests/integration/test_lists/test-db/l0_dgx_h100.yml
+++ b/tests/integration/test_lists/test-db/l0_dgx_h100.yml
@@ -78,6 +78,7 @@ l0_dgx_h100:
       auto_trigger: deepseek
   tests:
   - unittest/_torch/multi_gpu_modeling -k "deepseek"
+  - unittest/_torch/modules/test_fused_moe.py::test_fused_moe_alltoall[MNNVL]
   - unittest/_torch/modules/test_fused_moe.py::test_fused_moe_alltoall[DeepEP]
   - unittest/_torch/modules/test_fused_moe.py::test_fused_moe_alltoall[DeepEPLowLatency]
   - unittest/_torch/modules/test_fused_moe.py::test_fused_moe_w4afp8[MoEWeightLoadingMode.VANILLA-dtype0]
diff --git a/tests/unittest/_torch/modules/test_fused_moe.py b/tests/unittest/_torch/modules/test_fused_moe.py
index 397314bcab0..9a6720eca73 100644
--- a/tests/unittest/_torch/modules/test_fused_moe.py
+++ b/tests/unittest/_torch/modules/test_fused_moe.py
@@ -212,11 +212,14 @@ def per_rank_test_fused_moe_alltoall(job_id):
         weights = {}
         for expert_id in range(NUM_EXPERTS):
             w1_weight = torch.empty((INTERMEDIATE_SIZE, HIDDEN_SIZE),
-                                    dtype=dtype)
+                                    dtype=dtype,
+                                    device="cuda")
             w2_weight = torch.empty((HIDDEN_SIZE, INTERMEDIATE_SIZE),
-                                    dtype=dtype)
+                                    dtype=dtype,
+                                    device="cuda")
             w3_weight = torch.empty((INTERMEDIATE_SIZE, HIDDEN_SIZE),
-                                    dtype=dtype)
+                                    dtype=dtype,
+                                    device="cuda")
             torch.nn.init.xavier_uniform_(w1_weight)
             torch.nn.init.xavier_uniform_(w2_weight)
             torch.nn.init.xavier_uniform_(w3_weight)