diff --git a/tests/unittest/_torch/test_attention_no_cache.py b/tests/unittest/_torch/test_attention_no_cache.py
index 89de0147ae17..54e331a12025 100644
--- a/tests/unittest/_torch/test_attention_no_cache.py
+++ b/tests/unittest/_torch/test_attention_no_cache.py
@@ -182,6 +182,7 @@ def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
 
 # skip for blackwell
 @skip_blackwell
+@pytest.mark.skip(reason="https://nvbugspro.nvidia.com/bug/5247232")
 # Convert parameterized tests to pytest parametrize
 @pytest.mark.parametrize("accuracy", [(1e-2, 1e-3)],
                          ids=lambda x: f"atol={x[0]} rtol={x[1]}")