diff --git a/tensor2tensor/serving/query.py b/tensor2tensor/serving/query.py
index 69a7aadff..f64b30020 100644
--- a/tensor2tensor/serving/query.py
+++ b/tensor2tensor/serving/query.py
@@ -48,6 +48,8 @@
     "cloud_mlengine_model_version", None,
     "Version of the model to use. If None, requests will be "
     "sent to the default version.")
+flags.DEFINE_string("version", None, "Version of the model to use.")
+flags.DEFINE_string("version_label", None, "Label of the model to use.")
 
 
 def validate_flags():
@@ -72,7 +74,9 @@ def make_request_fn():
     request_fn = serving_utils.make_grpc_request_fn(
         servable_name=FLAGS.servable_name,
         server=FLAGS.server,
-        timeout_secs=FLAGS.timeout_secs)
+        timeout_secs=FLAGS.timeout_secs,
+        version_label=FLAGS.version_label,
+        version=FLAGS.version)
   return request_fn
 
 
diff --git a/tensor2tensor/serving/serving_utils.py b/tensor2tensor/serving/serving_utils.py
index a1b437282..f3ce7b596 100644
--- a/tensor2tensor/serving/serving_utils.py
+++ b/tensor2tensor/serving/serving_utils.py
@@ -105,7 +105,7 @@ def _decode(output_ids, output_decoder):
 
 
 
-def make_grpc_request_fn(servable_name, server, timeout_secs):
+def make_grpc_request_fn(servable_name, server, timeout_secs, version_label, version):
   """Wraps function to make grpc requests with runtime args."""
   stub = _create_stub(server)
 
@@ -113,6 +113,12 @@ def _make_grpc_request(examples):
     """Builds and sends request to TensorFlow model server."""
     request = predict_pb2.PredictRequest()
     request.model_spec.name = servable_name
+
+    if version_label is not None:
+        request.model_spec.version_label = version_label
+    elif version is not None:
+        request.model_spec.version = version
+
     request.inputs["input"].CopyFrom(
         tf.make_tensor_proto(
             [ex.SerializeToString() for ex in examples], shape=[len(examples)]))