diff --git a/domino/agents/_constants.py b/domino/agents/_constants.py
index 8a652aa6..889ef794 100644
--- a/domino/agents/_constants.py
+++ b/domino/agents/_constants.py
@@ -1,7 +1,7 @@
 # TODO update tag names to be prefixed with domino.agents in order to make origin clear
 EVALUATION_TAG_PREFIX = "domino.prog"
 DOMINO_INTERNAL_EVAL_TAG = "domino.internal.is_eval"
-MIN_MLFLOW_VERSION = "3.2.0"
-MIN_DOMINO_VERSION = "6.2.0"
+MIN_MLFLOW_VERSION = "3.9.0"
+MIN_DOMINO_VERSION = "6.3.0"
 LARGEST_MAX_RESULTS_PAGE_SIZE = 500
 EXPERIMENT_AGENT_TAG = "domino.agent"
diff --git a/setup.py b/setup.py
index 4a494a04..532a22a8 100644
--- a/setup.py
+++ b/setup.py
@@ -64,8 +64,8 @@ def get_version():
             "semver>=3.0.4",
             "pandas>=2.3.1",
             "numpy>=2.0.2",
-            "mlflow-skinny>=3.2.0,<3.3.0",
-            "mlflow-tracing>=3.2.0,<3.3.0",
+            "mlflow-skinny>=3.9.0,<3.10.0",
+            "mlflow-tracing>=3.9.0,<3.10.0",
         ],
         "dev": [
             "pytest-order>=1.3.0",
diff --git a/tests/integration/agents/test_tracing.py b/tests/integration/agents/test_tracing.py
index b3e8e512..0971e7cc 100644
--- a/tests/integration/agents/test_tracing.py
+++ b/tests/integration/agents/test_tracing.py
@@ -513,9 +513,12 @@ def parent2(x):
         assert sorted([(t.name, t.value) for trace in res.data for t in trace.evaluation_results if trace.name == "parent"]) \
                 == sorted([("mylabel", "category"), ("mymetric", 1.0)])
         assert len(span_data) == 4
-        assert sorted(span_data, key=lambda x: x[0]) == sorted([("parent", {'x':1, 'y': 2}, 3), \
-                ("parent2", {'x':1}, 1), ("unit_1", {'x':1}, 1), ("unit_2", {'x':2}, 2)
-        ], key=lambda x: x[0])
+
+        expected_spans = [("parent", {'x':1, 'y': 2}, 3), \
+                ("unit", {'x':1}, 1), ("unit", {'x':2}, 2), ("parent2", {'x':1}, 1)]
+
+        for e in expected_spans:
+                assert e in span_data
 
 def test_search_traces_time_filter_warning(setup_mlflow_tracking_server, tracing, mlflow, logging, caplog):
         """
@@ -555,8 +558,12 @@ def parent2(x):
 
         assert [trace.name for trace in res.data] == ["parent"]
         assert len(span_data) == 3
-        assert sorted(span_data, key=lambda x: x[0]) == sorted([("parent", {'x':1, 'y': 2}, 3), \
-                ("unit_1", {'x':1}, 1), ("unit_2", {'x':2}, 2)], key=lambda x: x[0])
+
+        expected_spans = [("parent", {'x':1, 'y': 2}, 3), \
+                ("unit", {'x':1}, 1), ("unit", {'x':2}, 2)]
+
+        for e in expected_spans:
+                assert e in span_data
 
 def test_search_traces_by_timestamp(setup_mlflow_tracking_server, mocker, mlflow, tracing, logging):
         @tracing.add_tracing(name="parent")