diff --git a/predict.py b/predict.py
index 3a1a2a0..f629a03 100644
--- a/predict.py
+++ b/predict.py
@@ -26,7 +26,7 @@ def parse_args():
     )
     parser.add_argument(
         "--benchmarks",
-        default="v5.2/live_benchmark_models.parquet",
+        default="v5.3/live_benchmark_models.parquet",
         help="Numerapi benchmark model path or local file.",
     )
     parser.add_argument("--model", required=True, help="Pickled model file or URL")
diff --git a/tests/test_predict.py b/tests/test_predict.py
index bc474a5..975539d 100644
--- a/tests/test_predict.py
+++ b/tests/test_predict.py
@@ -15,6 +15,13 @@ class TestPredict(unittest.TestCase):
     def setUp(self):
         pass
 
+    def test__parse_args_uses_default_data_versions(self):
+        with mock.patch("sys.argv", ["predict.py", "--model", "model.pkl"]):
+            args = predict.parse_args()
+
+        self.assertEqual(args.dataset, "v5.2/live.parquet")
+        self.assertEqual(args.benchmarks, "v5.3/live_benchmark_models.parquet")
+
     def test__is_python_version_pickle_error(self):
         self.assertTrue(
             predict.is_python_version_pickle_error(
@@ -81,7 +88,7 @@ def check_inputs(post_url, data, files):
     def test__main_exits_with_help_on_python_version_pickle_mismatch(self):
         args = argparse.Namespace(
             dataset="v5.2/live.parquet",
-            benchmarks="v5.2/live_benchmark_models.parquet",
+            benchmarks="v5.3/live_benchmark_models.parquet",
             model="model.pkl",
             output_dir="/tmp",
             post_url=None,