diff --git a/src/ntops/kernels/isinf.py b/src/ntops/kernels/isinf.py
new file mode 100644
index 0000000..9e5fdcb
--- /dev/null
+++ b/src/ntops/kernels/isinf.py
@@ -0,0 +1,17 @@
+import functools
+
+import ninetoothed
+from ninetoothed import Tensor
+
+from ntops.kernels.element_wise import arrangement
+
+
+def application(input, output):
+    pos_result = input == float("+inf")
+    neg_result = input == float("-inf")
+    output = pos_result or neg_result  # noqa: F841
+
+
+@functools.cache
+def make(ndim):
+    return ninetoothed.make(arrangement, application, (Tensor(ndim), Tensor(ndim)))
diff --git a/src/ntops/torch.py b/src/ntops/torch.py
index 9a4c895..41ae3f2 100644
--- a/src/ntops/torch.py
+++ b/src/ntops/torch.py
@@ -8,6 +8,7 @@
 import ntops.kernels.div
 import ntops.kernels.exp
 import ntops.kernels.gelu
+import ntops.kernels.isinf
 import ntops.kernels.mm
 import ntops.kernels.mul
 import ntops.kernels.relu
@@ -109,6 +110,16 @@ def gelu(input, approximate="none"):
     return output
 
 
+def isinf(input):
+    output = torch.empty_like(input)
+
+    kernel = ntops.kernels.isinf.make(input.ndim)
+
+    kernel(input, output)
+
+    return output
+
+
 def mm(input, mat2, *, out=None):
     m, _ = input.shape
     _, n = mat2.shape
diff --git a/tests/test_isinf.py b/tests/test_isinf.py
new file mode 100644
index 0000000..c7739db
--- /dev/null
+++ b/tests/test_isinf.py
@@ -0,0 +1,32 @@
+import pytest
+import torch
+
+import ntops.torch
+from tests.skippers import skip_if_cuda_not_available
+from tests.utils import generate_arguments
+
+
+@skip_if_cuda_not_available
+@pytest.mark.parametrize(*generate_arguments())
+def test_cuda(shape, dtype, atol, rtol):
+    device = "cuda"
+
+    def generate_inf_tensor(shape, dtype, device):
+        x = torch.randn(shape, dtype=dtype, device=device)
+
+        probs = (0.2, 0.6)
+        prob_tensor = torch.rand(shape, device=device)
+
+        mask = (probs[0] < prob_tensor) & (prob_tensor < probs[1])
+        x[mask] = float("inf")
+        mask = probs[1] < prob_tensor
+        x[mask] = float("-inf")
+
+        return x
+
+    input = generate_inf_tensor(shape, dtype, device)
+
+    ninetoothed_output = ntops.torch.isinf(input)
+    reference_output = torch.isinf(input)
+
+    assert torch.equal(ninetoothed_output, reference_output)