diff --git a/tests/integration/train_tests.py b/tests/integration/train_tests.py
index 2e2727d65d..2511fbe463 100644
--- a/tests/integration/train_tests.py
+++ b/tests/integration/train_tests.py
@@ -505,6 +505,7 @@ def test_tpu_zero1_gradient_accumulation(self):
   @pytest.mark.integration_test
   @pytest.mark.gpu_only
   @pytest.mark.scheduled_only
+  @pytest.mark.skip(reason="b/489133823. Previously transient in b/462548581.")
   def test_gpu_zero1_gradient_accumulation(self):
     os.environ["NVTE_FUSED_ATTN"] = "1"  # Enable fused attention
     zero1_ga = [  # tests Zero-1 optimizer sharding with gradient accumulation