diff --git a/test/e2e/helm/test_kserve_sklearn.py b/test/e2e/helm/test_kserve_sklearn.py
index 5b8d08d992a..140c34b5428 100644
--- a/test/e2e/helm/test_kserve_sklearn.py
+++ b/test/e2e/helm/test_kserve_sklearn.py
@@ -28,12 +28,13 @@
     constants
 )
 
-from ..common.utils import KSERVE_TEST_NAMESPACE, predict
+from ..common.utils import KSERVE_TEST_NAMESPACE, predict, get_cluster_ip
 
 
 @pytest.mark.helm
 def test_sklearn_kserve():
     service_name = "isvc-sklearn-helm"
+    protocol_version = "v2"
 
     predictor = V1beta1PredictorSpec(
         min_replicas=1,
@@ -43,7 +44,7 @@ def test_sklearn_kserve():
             ),
             runtime="kserve-mlserver",
             storage_uri="gs://seldon-models/sklearn/mms/lr_model",
-            protocol_version="v2",
+            protocol_version=protocol_version,
             resources=V1ResourceRequirements(
                 requests={"cpu": "50m", "memory": "128Mi"},
                 limits={"cpu": "100m", "memory": "512Mi"},
@@ -65,6 +66,14 @@ def test_sklearn_kserve():
     kserve_client.create(isvc)
     kserve_client.wait_isvc_ready(
         service_name, namespace=KSERVE_TEST_NAMESPACE)
+    kserve_client.wait_model_ready(
+        service_name,
+        model_name=service_name,
+        isvc_namespace=KSERVE_TEST_NAMESPACE,
+        isvc_version=constants.KSERVE_V1BETA1_VERSION,
+        protocol_version=protocol_version,
+        cluster_ip=get_cluster_ip(),
+    )
 
     res = predict(service_name, "./data/iris_input_v2.json",
                   protocol_version="v2")
diff --git a/test/e2e/predictor/test_lightgbm.py b/test/e2e/predictor/test_lightgbm.py
index 0396a9827e7..77a2eead173 100644
--- a/test/e2e/predictor/test_lightgbm.py
+++ b/test/e2e/predictor/test_lightgbm.py
@@ -25,7 +25,7 @@
                     V1beta1ModelFormat, V1beta1ModelSpec, V1beta1PredictorSpec,
                     constants)
 
-from ..common.utils import KSERVE_TEST_NAMESPACE, predict, predict_grpc
+from ..common.utils import KSERVE_TEST_NAMESPACE, predict, predict_grpc, get_cluster_ip
 
 
 @pytest.mark.fast
@@ -104,6 +104,7 @@ def test_lightgbm_runtime_kserve():
 @pytest.mark.fast
 def test_lightgbm_v2_runtime_mlserver():
     service_name = "isvc-lightgbm-v2-runtime"
+    protocol_version = "v2"
 
     predictor = V1beta1PredictorSpec(
         min_replicas=1,
@@ -113,7 +114,7 @@ def test_lightgbm_v2_runtime_mlserver():
             ),
             runtime="kserve-mlserver",
             storage_uri="gs://kfserving-examples/models/lightgbm/v2/iris",
-            protocol_version="v2",
+            protocol_version=protocol_version,
             resources=V1ResourceRequirements(
                 requests={"cpu": "50m", "memory": "128Mi"},
                 limits={"cpu": "1", "memory": "1Gi"},
@@ -135,6 +136,14 @@ def test_lightgbm_v2_runtime_mlserver():
     kserve_client.create(isvc)
     kserve_client.wait_isvc_ready(
         service_name, namespace=KSERVE_TEST_NAMESPACE)
+    kserve_client.wait_model_ready(
+        service_name,
+        model_name=service_name,
+        isvc_namespace=KSERVE_TEST_NAMESPACE,
+        isvc_version=constants.KSERVE_V1BETA1_VERSION,
+        protocol_version=protocol_version,
+        cluster_ip=get_cluster_ip(),
+    )
 
     res = predict(service_name, "./data/iris_input_v2.json",
                   protocol_version="v2")
diff --git a/test/e2e/predictor/test_sklearn.py b/test/e2e/predictor/test_sklearn.py
index 94dd54ce399..ffd37e32e17 100644
--- a/test/e2e/predictor/test_sklearn.py
+++ b/test/e2e/predictor/test_sklearn.py
@@ -25,7 +25,7 @@
 
 import kserve.protocol.grpc.grpc_predict_v2_pb2 as inference_pb2
 
-from ..common.utils import KSERVE_TEST_NAMESPACE, predict, predict_grpc
+from ..common.utils import KSERVE_TEST_NAMESPACE, predict, predict_grpc, get_cluster_ip
 
 
 @pytest.mark.slow
@@ -62,11 +62,13 @@ def test_sklearn_kserve():
 @pytest.mark.slow
 def test_sklearn_v2_mlserver():
     service_name = "sklearn-v2-mlserver"
+    protocol_version = "v2"
+
     predictor = V1beta1PredictorSpec(
         min_replicas=1,
         sklearn=V1beta1SKLearnSpec(
             storage_uri="gs://seldon-models/sklearn/mms/lr_model",
-            protocol_version="v2",
+            protocol_version=protocol_version,
             resources=V1ResourceRequirements(
                 requests={"cpu": "50m", "memory": "128Mi"},
                 limits={"cpu": "100m", "memory": "512Mi"},
@@ -86,6 +88,14 @@ def test_sklearn_v2_mlserver():
     kserve_client = KServeClient(config_file=os.environ.get("KUBECONFIG", "~/.kube/config"))
     kserve_client.create(isvc)
     kserve_client.wait_isvc_ready(service_name, namespace=KSERVE_TEST_NAMESPACE)
+    kserve_client.wait_model_ready(
+        service_name,
+        model_name=service_name,
+        isvc_namespace=KSERVE_TEST_NAMESPACE,
+        isvc_version=constants.KSERVE_V1BETA1_VERSION,
+        protocol_version=protocol_version,
+        cluster_ip=get_cluster_ip(),
+    )
 
     res = predict(service_name, "./data/iris_input_v2.json", protocol_version="v2")
     assert res["outputs"][0]["data"] == [1, 1]
@@ -131,6 +141,7 @@ def test_sklearn_runtime_kserve():
 @pytest.mark.slow
 def test_sklearn_v2_runtime_mlserver():
     service_name = "isvc-sklearn-v2-runtime"
+    protocol_version = "v2"
 
     predictor = V1beta1PredictorSpec(
         min_replicas=1,
@@ -140,7 +151,7 @@ def test_sklearn_v2_runtime_mlserver():
             ),
             runtime="kserve-mlserver",
             storage_uri="gs://seldon-models/sklearn/mms/lr_model",
-            protocol_version="v2",
+            protocol_version=protocol_version,
             resources=V1ResourceRequirements(
                 requests={"cpu": "50m", "memory": "128Mi"},
                 limits={"cpu": "100m", "memory": "512Mi"},
@@ -160,6 +171,14 @@ def test_sklearn_v2_runtime_mlserver():
     kserve_client = KServeClient(config_file=os.environ.get("KUBECONFIG", "~/.kube/config"))
     kserve_client.create(isvc)
     kserve_client.wait_isvc_ready(service_name, namespace=KSERVE_TEST_NAMESPACE)
+    kserve_client.wait_model_ready(
+        service_name,
+        model_name=service_name,
+        isvc_namespace=KSERVE_TEST_NAMESPACE,
+        isvc_version=constants.KSERVE_V1BETA1_VERSION,
+        protocol_version=protocol_version,
+        cluster_ip=get_cluster_ip(),
+    )
 
     res = predict(service_name, "./data/iris_input_v2.json", protocol_version="v2")
     assert res["outputs"][0]["data"] == [1, 1]
diff --git a/test/e2e/predictor/test_xgboost.py b/test/e2e/predictor/test_xgboost.py
index 1f92ef85beb..f6f853beae7 100644
--- a/test/e2e/predictor/test_xgboost.py
+++ b/test/e2e/predictor/test_xgboost.py
@@ -23,7 +23,7 @@
                     V1beta1ModelSpec, V1beta1PredictorSpec, V1beta1XGBoostSpec,
                     constants)
 
-from ..common.utils import KSERVE_TEST_NAMESPACE, predict, predict_grpc
+from ..common.utils import KSERVE_TEST_NAMESPACE, predict, predict_grpc, get_cluster_ip
 
 
 @pytest.mark.fast
@@ -62,12 +62,14 @@ def test_xgboost_kserve():
 @pytest.mark.fast
 def test_xgboost_v2_mlserver():
     service_name = "isvc-xgboost-v2-mlserver"
+    protocol_version = "v2"
+
     predictor = V1beta1PredictorSpec(
         min_replicas=1,
         xgboost=V1beta1XGBoostSpec(
             storage_uri="gs://kfserving-examples/models/xgboost/iris",
             env=[V1EnvVar(name="MLSERVER_MODEL_PARALLEL_WORKERS", value="0")],
-            protocol_version="v2",
+            protocol_version=protocol_version,
             resources=V1ResourceRequirements(
                 requests={"cpu": "50m", "memory": "128Mi"},
                 limits={"cpu": "100m", "memory": "1024Mi"},
@@ -89,6 +91,14 @@ def test_xgboost_v2_mlserver():
     kserve_client.create(isvc)
     kserve_client.wait_isvc_ready(
         service_name, namespace=KSERVE_TEST_NAMESPACE)
+    kserve_client.wait_model_ready(
+        service_name,
+        model_name=service_name,
+        isvc_namespace=KSERVE_TEST_NAMESPACE,
+        isvc_version=constants.KSERVE_V1BETA1_VERSION,
+        protocol_version=protocol_version,
+        cluster_ip=get_cluster_ip(),
+    )
 
     res = predict(service_name, "./data/iris_input_v2.json",
                   protocol_version="v2")
@@ -136,6 +146,7 @@ def test_xgboost_runtime_kserve():
 @pytest.mark.fast
 def test_xgboost_v2_runtime_mlserver():
     service_name = "isvc-xgboost-v2-runtime"
+    protocol_version = "v2"
 
     predictor = V1beta1PredictorSpec(
         min_replicas=1,
@@ -145,7 +156,7 @@ def test_xgboost_v2_runtime_mlserver():
             ),
             runtime="kserve-mlserver",
             storage_uri="gs://kfserving-examples/models/xgboost/iris",
-            protocol_version="v2",
+            protocol_version=protocol_version,
             resources=V1ResourceRequirements(
                 requests={"cpu": "50m", "memory": "128Mi"},
                 limits={"cpu": "100m", "memory": "1024Mi"},
@@ -167,6 +178,14 @@ def test_xgboost_v2_runtime_mlserver():
     kserve_client.create(isvc)
     kserve_client.wait_isvc_ready(
         service_name, namespace=KSERVE_TEST_NAMESPACE)
+    kserve_client.wait_model_ready(
+        service_name,
+        model_name=service_name,
+        isvc_namespace=KSERVE_TEST_NAMESPACE,
+        isvc_version=constants.KSERVE_V1BETA1_VERSION,
+        protocol_version=protocol_version,
+        cluster_ip=get_cluster_ip(),
+    )
 
     res = predict(service_name, "./data/iris_input_v2.json",
                   protocol_version="v2")