diff --git a/pytest.ini b/pytest.ini
index b9bb2d26ca..8d456e23ba 100644
--- a/pytest.ini
+++ b/pytest.ini
@@ -4,3 +4,4 @@ testpaths = test/
 python_paths = ./
 markers =
     gpu_test: marks cuda tests
+    npu_test: marks ascend npu tests
diff --git a/test/torchtext_unittest/models/npu_tests/models_npu_test.py b/test/torchtext_unittest/models/npu_tests/models_npu_test.py
new file mode 100644
index 0000000000..f3ff4919b2
--- /dev/null
+++ b/test/torchtext_unittest/models/npu_tests/models_npu_test.py
@@ -0,0 +1,33 @@
+import importlib
+import unittest
+
+import pytest
+import torch
+from torchtext_unittest.common.torchtext_test_case import TorchtextTestCase
+from torchtext_unittest.models.roberta_models_test_impl import RobertaBaseTestModels
+from torchtext_unittest.models.t5_models_test_impl import T5BaseTestModels
+
+
+def is_npu_available(check_device=False):
+    "Checks if `torch_npu` is installed and potentially if a NPU is in the environment"
+    if importlib.util.find_spec("torch") is None or importlib.util.find_spec("torch_npu") is None:
+        return False
+
+    import torch
+    import torch_npu  # noqa: F401
+
+    if check_device:
+        try:
+            # Will raise a RuntimeError if no NPU is found
+            _ = torch.npu.device_count()
+            return torch.npu.is_available()
+        except RuntimeError:
+            return False
+    return hasattr(torch, "npu") and torch.npu.is_available()
+ 
+
+@pytest.mark.npu_test
+@unittest.skipIf(not is_npu_available(), reason="Ascend NPU is not available")
+class TestModels32NPU(RobertaBaseTestModels, T5BaseTestModels, TorchtextTestCase):
+    dtype = torch.float32
+    device = torch.device("npu")