diff --git a/Orange/classification/xgb.py b/Orange/classification/xgb.py
index cd67376513d..48ca446c3f7 100644
--- a/Orange/classification/xgb.py
+++ b/Orange/classification/xgb.py
@@ -1,6 +1,5 @@
 # pylint: disable=too-many-arguments
 from typing import Tuple
-import sys
 
 import numpy as np
 
@@ -13,9 +12,6 @@
 
 __all__ = ["XGBClassifier", "XGBRFClassifier"]
 
-# Temporary workaround for https://github.com/dmlc/xgboost/issues/7156
-N_JOBS = 1 if sys.platform == "darwin" else None
-
 
 class _FeatureScorerMixin(LearnerScorer):
     feature_type = Variable
@@ -37,7 +33,7 @@ def __init__(self,
                  objective="binary:logistic",
                  booster=None,
                  tree_method=None,
-                 n_jobs=N_JOBS,
+                 n_jobs=None,
                  gamma=None,
                  min_child_weight=None,
                  max_delta_step=None,
@@ -100,7 +96,7 @@ def __init__(self,
                  objective="binary:logistic",
                  booster=None,
                  tree_method=None,
-                 n_jobs=N_JOBS,
+                 n_jobs=None,
                  gamma=None,
                  min_child_weight=None,
                  max_delta_step=None,
diff --git a/Orange/modelling/tests/test_xgb.py b/Orange/modelling/tests/test_xgb.py
index 8ee926e5aed..761dc961775 100644
--- a/Orange/modelling/tests/test_xgb.py
+++ b/Orange/modelling/tests/test_xgb.py
@@ -1,6 +1,5 @@
 import unittest
 from typing import Callable, Union
-import xgboost
 
 from Orange.data import Table
 from Orange.evaluation import CrossValidation
@@ -26,14 +25,6 @@ def setUpClass(cls):
         cls.iris = Table("iris")
         cls.housing = Table("housing")
 
-    def test_xgboost_workaround(self):
-        self.assertLessEqual(
-            list(map(int, xgboost.__version__.split("."))),
-            [1, 5, 0],
-            "\nCheck the new release of xgboost and revert this "
-            "commit if possible, or do something else if not."
-        )
-
     @test_learners
     def test_cls(self, learner_class: Union[XGBLearner, XGBRFLearner]):
         booster = learner_class()
diff --git a/Orange/regression/xgb.py b/Orange/regression/xgb.py
index 5cce529dc2f..4c8f4b7d362 100644
--- a/Orange/regression/xgb.py
+++ b/Orange/regression/xgb.py
@@ -1,6 +1,5 @@
 # pylint: disable=too-many-arguments
 from typing import Tuple
-import sys
 
 import numpy as np
 import xgboost
@@ -12,9 +11,6 @@
 
 __all__ = ["XGBRegressor", "XGBRFRegressor"]
 
-# Temporary workaround for https://github.com/dmlc/xgboost/issues/7156
-N_JOBS = 1 if sys.platform == "darwin" else None
-
 
 class _FeatureScorerMixin(LearnerScorer):
     feature_type = Variable
@@ -36,7 +32,7 @@ def __init__(self,
                  objective="reg:squarederror",
                  booster=None,
                  tree_method=None,
-                 n_jobs=N_JOBS,
+                 n_jobs=None,
                  gamma=None,
                  min_child_weight=None,
                  max_delta_step=None,
@@ -88,7 +84,7 @@ def __init__(self,
                  objective="reg:squarederror",
                  booster=None,
                  tree_method=None,
-                 n_jobs=N_JOBS,
+                 n_jobs=None,
                  gamma=None,
                  min_child_weight=None,
                  max_delta_step=None,