diff --git a/.github/workflows/benchmark.yml b/.github/workflows/benchmark.yml
index 3ae7aa91d644..44c10172ce64 100644
--- a/.github/workflows/benchmark.yml
+++ b/.github/workflows/benchmark.yml
@@ -107,4 +107,4 @@ jobs:
 
       - name: Run non-benchmark tests
         working-directory: py-polars
-        run: pytest -m 'not release' -n auto --dist loadgroup
+        run: pytest -m 'not release and not debug' -n auto --dist loadgroup
diff --git a/py-polars/pyproject.toml b/py-polars/pyproject.toml
index db8d04a78608..e0b9c7cae7d0 100644
--- a/py-polars/pyproject.toml
+++ b/py-polars/pyproject.toml
@@ -211,6 +211,7 @@ markers = [
   "write_disk: Tests that write to disk",
   "slow: Tests with a longer than average runtime.",
   "release: Tests that should be run on a Polars release build.",
+  "debug: Tests that should be run on a Polars debug build.",
   "docs: Documentation code snippets",
 ]
 filterwarnings = [
diff --git a/py-polars/tests/unit/streaming/test_streaming_sort.py b/py-polars/tests/unit/streaming/test_streaming_sort.py
index e61311459e0c..e30859b74d3c 100644
--- a/py-polars/tests/unit/streaming/test_streaming_sort.py
+++ b/py-polars/tests/unit/streaming/test_streaming_sort.py
@@ -92,6 +92,7 @@ def test_ooc_sort(tmp_path: Path, monkeypatch: Any) -> None:
         assert_series_equal(out, s.sort(descending=descending))
 
 
+@pytest.mark.debug()
 @pytest.mark.write_disk()
 @pytest.mark.parametrize("spill_source", [True, False])
 def test_streaming_sort(
diff --git a/py-polars/tests/unit/test_cse.py b/py-polars/tests/unit/test_cse.py
index b12b5f3fab18..70ca7e5c8185 100644
--- a/py-polars/tests/unit/test_cse.py
+++ b/py-polars/tests/unit/test_cse.py
@@ -172,6 +172,7 @@ def test_schema_row_index_cse() -> None:
     assert_frame_equal(result, expected)
 
 
+@pytest.mark.debug()
 def test_cse_expr_selection_context(monkeypatch: Any, capfd: Any) -> None:
     monkeypatch.setenv("POLARS_VERBOSE", "1")
     q = pl.LazyFrame(