diff --git a/neuralnetworks/1.3/Android.bp b/neuralnetworks/1.3/Android.bp
index 8e3e9f1..0b07a58 100644
--- a/neuralnetworks/1.3/Android.bp
+++ b/neuralnetworks/1.3/Android.bp
@@ -9,6 +9,7 @@
     srcs: [
         "types.hal",
         "IDevice.hal",
+        "IPreparedModel.hal",
         "IPreparedModelCallback.hal",
     ],
     interfaces: [
diff --git a/neuralnetworks/1.3/IPreparedModel.hal b/neuralnetworks/1.3/IPreparedModel.hal
new file mode 100644
index 0000000..c04809f
--- /dev/null
+++ b/neuralnetworks/1.3/IPreparedModel.hal
@@ -0,0 +1,90 @@
+/*
+ * Copyright (C) 2019 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package android.hardware.neuralnetworks@1.3;
+
+import @1.0::ErrorStatus;
+import @1.0::Request;
+import @1.2::MeasureTiming;
+import @1.2::IExecutionCallback;
+import @1.2::IPreparedModel;
+
+/**
+ * IPreparedModel describes a model that has been prepared for execution and
+ * is used to launch executions.
+ */
+interface IPreparedModel extends @1.2::IPreparedModel {
+    /**
+     * Launches an asynchronous execution on a prepared model.
+     *
+     * The execution is performed asynchronously with respect to the caller.
+     * execute_1_3 must verify the inputs to the function are correct. If there is
+     * an error, execute_1_3 must immediately invoke the callback with the
+     * appropriate ErrorStatus value, then return with the same ErrorStatus. If
+     * the inputs to the function are valid and there is no error, execute_1_3 must
+     * launch an asynchronous task to perform the execution in the background,
+     * and immediately return with ErrorStatus::NONE. If the asynchronous task
+     * fails to launch, execute_1_3 must immediately invoke the callback with
+     * ErrorStatus::GENERAL_FAILURE, then return with
+     * ErrorStatus::GENERAL_FAILURE.
+     *
+     * When the asynchronous task has finished its execution, it must
+     * immediately invoke the callback object provided as an input to the
+     * execute_1_3 function. This callback must be provided with the ErrorStatus of
+     * the execution.
+     *
+     * If the launch is successful, the caller must not change the content of
+     * any data object referenced by 'request' (described by the
+     * {@link @1.0::DataLocation} of a {@link @1.0::RequestArgument}) until the
+     * asynchronous task has invoked the callback object. The asynchronous task
+     * must not change the content of any of the data objects corresponding to
+     * 'request' inputs.
+     *
+     * If the prepared model was prepared from a model wherein all tensor
+     * operands have fully specified dimensions, and the inputs to the function
+     * are valid, then:
+     * - the execution should launch successfully (ErrorStatus::NONE): There
+     *   must be no failure unless the device itself is in a bad state.
+     * - if at execution time every operation's input operands have legal
+     *   values, the execution should complete successfully (ErrorStatus::NONE):
+     *   There must be no failure unless the device itself is in a bad state.
+     *
+     * Any number of calls to the execute, execute_1_2, execute_1_3, and executeSynchronously
+     * functions, in any combination, may be made concurrently, even on the same
+     * IPreparedModel object.
+     *
+     * @param request The input and output information on which the prepared
+     *                model is to be executed.
+     * @param measure Specifies whether or not to measure duration of the execution.
+     *                The duration runs from the time the driver sees the call
+     *                to the execute_1_3 function to the time the driver invokes
+     *                the callback.
+     * @param callback A callback object used to return the error status of
+     *                 the execution. The callback object's notify function must
+     *                 be called exactly once, even if the execution was
+     *                 unsuccessful.
+     * @return status Error status of the call, must be:
+     *                - NONE if task is successfully launched
+     *                - DEVICE_UNAVAILABLE if driver is offline or busy
+     *                - GENERAL_FAILURE if there is an unspecified error
+     *                - OUTPUT_INSUFFICIENT_SIZE if provided output buffer is
+     *                  not large enough to store the resultant values
+     *                - INVALID_ARGUMENT if one of the input arguments is
+     *                  invalid
+     */
+    execute_1_3(Request request, MeasureTiming measure, IExecutionCallback callback)
+        generates (ErrorStatus status);
+};
diff --git a/neuralnetworks/1.3/IPreparedModelCallback.hal b/neuralnetworks/1.3/IPreparedModelCallback.hal
index 7cc5ae0..ff295a2 100644
--- a/neuralnetworks/1.3/IPreparedModelCallback.hal
+++ b/neuralnetworks/1.3/IPreparedModelCallback.hal
@@ -18,7 +18,7 @@
 
 import @1.0::ErrorStatus;
 import @1.2::IPreparedModelCallback;
-import @1.2::IPreparedModel;
+import IPreparedModel;
 
 /**
  * IPreparedModelCallback must be used to return a prepared model produced by an
diff --git a/neuralnetworks/1.3/vts/functional/Callbacks.cpp b/neuralnetworks/1.3/vts/functional/Callbacks.cpp
index 435db46..4f08e72 100644
--- a/neuralnetworks/1.3/vts/functional/Callbacks.cpp
+++ b/neuralnetworks/1.3/vts/functional/Callbacks.cpp
@@ -54,7 +54,7 @@
 }
 
 Return<void> PreparedModelCallback::notify_1_3(ErrorStatus errorStatus,
-                                               const sp<V1_2::IPreparedModel>& preparedModel) {
+                                               const sp<V1_3::IPreparedModel>& preparedModel) {
     return notify(errorStatus, preparedModel);
 }
 
diff --git a/neuralnetworks/1.3/vts/functional/CompilationCachingTests.cpp b/neuralnetworks/1.3/vts/functional/CompilationCachingTests.cpp
index ea2398b..d8a7534 100644
--- a/neuralnetworks/1.3/vts/functional/CompilationCachingTests.cpp
+++ b/neuralnetworks/1.3/vts/functional/CompilationCachingTests.cpp
@@ -52,7 +52,6 @@
 using V1_0::ErrorStatus;
 using V1_1::ExecutionPreference;
 using V1_2::Constant;
-using V1_2::IPreparedModel;
 using V1_2::OperationType;
 
 namespace float32_model {
diff --git a/neuralnetworks/1.3/vts/functional/GeneratedTestHarness.cpp b/neuralnetworks/1.3/vts/functional/GeneratedTestHarness.cpp
index 4f9b6f9..325d641 100644
--- a/neuralnetworks/1.3/vts/functional/GeneratedTestHarness.cpp
+++ b/neuralnetworks/1.3/vts/functional/GeneratedTestHarness.cpp
@@ -29,6 +29,7 @@
 #include <android/hardware/neuralnetworks/1.2/IPreparedModelCallback.h>
 #include <android/hardware/neuralnetworks/1.2/types.h>
 #include <android/hardware/neuralnetworks/1.3/IDevice.h>
+#include <android/hardware/neuralnetworks/1.3/IPreparedModel.h>
 #include <android/hardware/neuralnetworks/1.3/IPreparedModelCallback.h>
 #include <android/hardware/neuralnetworks/1.3/types.h>
 #include <android/hidl/allocator/1.0/IAllocator.h>
@@ -61,7 +62,6 @@
 using V1_0::Request;
 using V1_1::ExecutionPreference;
 using V1_2::Constant;
-using V1_2::IPreparedModel;
 using V1_2::MeasureTiming;
 using V1_2::OperationType;
 using V1_2::OutputShape;
@@ -181,7 +181,7 @@
 static Return<ErrorStatus> ExecutePreparedModel(const sp<IPreparedModel>& preparedModel,
                                                 const Request& request, MeasureTiming measure,
                                                 sp<ExecutionCallback>& callback) {
-    return preparedModel->execute_1_2(request, measure, callback);
+    return preparedModel->execute_1_3(request, measure, callback);
 }
 static Return<ErrorStatus> ExecutePreparedModel(const sp<IPreparedModel>& preparedModel,
                                                 const Request& request, MeasureTiming measure,
diff --git a/neuralnetworks/1.3/vts/functional/GeneratedTestHarness.h b/neuralnetworks/1.3/vts/functional/GeneratedTestHarness.h
index b9277cf..45cff5b 100644
--- a/neuralnetworks/1.3/vts/functional/GeneratedTestHarness.h
+++ b/neuralnetworks/1.3/vts/functional/GeneratedTestHarness.h
@@ -17,8 +17,8 @@
 #ifndef ANDROID_HARDWARE_NEURALNETWORKS_V1_3_GENERATED_TEST_HARNESS_H
 #define ANDROID_HARDWARE_NEURALNETWORKS_V1_3_GENERATED_TEST_HARNESS_H
 
-#include <android/hardware/neuralnetworks/1.2/IPreparedModel.h>
 #include <android/hardware/neuralnetworks/1.3/IDevice.h>
+#include <android/hardware/neuralnetworks/1.3/IPreparedModel.h>
 #include <android/hardware/neuralnetworks/1.3/types.h>
 #include <functional>
 #include <vector>
@@ -55,10 +55,9 @@
 
 Model createModel(const test_helper::TestModel& testModel);
 
-void PrepareModel(const sp<IDevice>& device, const Model& model,
-                  sp<V1_2::IPreparedModel>* preparedModel);
+void PrepareModel(const sp<IDevice>& device, const Model& model, sp<IPreparedModel>* preparedModel);
 
-void EvaluatePreparedModel(const sp<V1_2::IPreparedModel>& preparedModel,
+void EvaluatePreparedModel(const sp<IPreparedModel>& preparedModel,
                            const test_helper::TestModel& testModel, bool testDynamicOutputShape);
 
 }  // namespace android::hardware::neuralnetworks::V1_3::vts::functional
diff --git a/neuralnetworks/1.3/vts/functional/ValidateBurst.cpp b/neuralnetworks/1.3/vts/functional/ValidateBurst.cpp
index 2c97294..7df8046 100644
--- a/neuralnetworks/1.3/vts/functional/ValidateBurst.cpp
+++ b/neuralnetworks/1.3/vts/functional/ValidateBurst.cpp
@@ -40,7 +40,6 @@
 using V1_2::FmqResultDatum;
 using V1_2::IBurstCallback;
 using V1_2::IBurstContext;
-using V1_2::IPreparedModel;
 using V1_2::MeasureTiming;
 using V1_2::Timing;
 using ExecutionBurstCallback = ExecutionBurstController::ExecutionBurstCallback;
diff --git a/neuralnetworks/1.3/vts/functional/ValidateModel.cpp b/neuralnetworks/1.3/vts/functional/ValidateModel.cpp
index bdda790..1ff02dc 100644
--- a/neuralnetworks/1.3/vts/functional/ValidateModel.cpp
+++ b/neuralnetworks/1.3/vts/functional/ValidateModel.cpp
@@ -27,7 +27,6 @@
 using V1_0::ErrorStatus;
 using V1_0::OperandLifeTime;
 using V1_1::ExecutionPreference;
-using V1_2::IPreparedModel;
 using V1_2::OperationType;
 using V1_2::OperationTypeRange;
 using V1_2::SymmPerChannelQuantParams;
@@ -61,7 +60,7 @@
     preparedModelCallback->wait();
     ErrorStatus prepareReturnStatus = preparedModelCallback->getStatus();
     ASSERT_EQ(ErrorStatus::INVALID_ARGUMENT, prepareReturnStatus);
-    sp<IPreparedModel> preparedModel = getPreparedModel_1_2(preparedModelCallback);
+    sp<IPreparedModel> preparedModel = getPreparedModel_1_3(preparedModelCallback);
     ASSERT_EQ(nullptr, preparedModel.get());
 }
 
diff --git a/neuralnetworks/1.3/vts/functional/ValidateRequest.cpp b/neuralnetworks/1.3/vts/functional/ValidateRequest.cpp
index c00512c..2cf30d5 100644
--- a/neuralnetworks/1.3/vts/functional/ValidateRequest.cpp
+++ b/neuralnetworks/1.3/vts/functional/ValidateRequest.cpp
@@ -29,7 +29,6 @@
 
 using V1_0::ErrorStatus;
 using V1_0::Request;
-using V1_2::IPreparedModel;
 using V1_2::MeasureTiming;
 using V1_2::OutputShape;
 using V1_2::Timing;
@@ -61,11 +60,11 @@
 
     // asynchronous
     {
-        SCOPED_TRACE(message + " [execute_1_2]");
+        SCOPED_TRACE(message + " [execute_1_3]");
 
         sp<ExecutionCallback> executionCallback = new ExecutionCallback();
         Return<ErrorStatus> executeLaunchStatus =
-                preparedModel->execute_1_2(request, measure, executionCallback);
+                preparedModel->execute_1_3(request, measure, executionCallback);
         ASSERT_TRUE(executeLaunchStatus.isOk());
         ASSERT_EQ(ErrorStatus::INVALID_ARGUMENT, static_cast<ErrorStatus>(executeLaunchStatus));
 
diff --git a/neuralnetworks/1.3/vts/functional/VtsHalNeuralnetworks.cpp b/neuralnetworks/1.3/vts/functional/VtsHalNeuralnetworks.cpp
index da5d95b..625913d 100644
--- a/neuralnetworks/1.3/vts/functional/VtsHalNeuralnetworks.cpp
+++ b/neuralnetworks/1.3/vts/functional/VtsHalNeuralnetworks.cpp
@@ -34,7 +34,6 @@
 using V1_0::ErrorStatus;
 using V1_0::Request;
 using V1_1::ExecutionPreference;
-using V1_2::IPreparedModel;
 
 // internal helper function
 void createPreparedModel(const sp<IDevice>& device, const Model& model,
@@ -64,7 +63,7 @@
     // retrieve prepared model
     preparedModelCallback->wait();
     const ErrorStatus prepareReturnStatus = preparedModelCallback->getStatus();
-    *preparedModel = getPreparedModel_1_2(preparedModelCallback);
+    *preparedModel = getPreparedModel_1_3(preparedModelCallback);
 
     // The getSupportedOperations_1_3 call returns a list of operations that are
     // guaranteed not to fail if prepareModel_1_3 is called, and
@@ -165,7 +164,7 @@
 
 INSTANTIATE_GENERATED_TEST(ValidationTest, [](const test_helper::TestModel&) { return true; });
 
-sp<IPreparedModel> getPreparedModel_1_2(const sp<PreparedModelCallback>& callback) {
+sp<IPreparedModel> getPreparedModel_1_3(const sp<PreparedModelCallback>& callback) {
     sp<V1_0::IPreparedModel> preparedModelV1_0 = callback->getPreparedModel();
     return IPreparedModel::castFrom(preparedModelV1_0).withDefault(nullptr);
 }
diff --git a/neuralnetworks/1.3/vts/functional/VtsHalNeuralnetworks.h b/neuralnetworks/1.3/vts/functional/VtsHalNeuralnetworks.h
index 9ccc911..8cb42d4 100644
--- a/neuralnetworks/1.3/vts/functional/VtsHalNeuralnetworks.h
+++ b/neuralnetworks/1.3/vts/functional/VtsHalNeuralnetworks.h
@@ -17,8 +17,8 @@
 #ifndef ANDROID_HARDWARE_NEURALNETWORKS_V1_3_VTS_HAL_NEURALNETWORKS_H
 #define ANDROID_HARDWARE_NEURALNETWORKS_V1_3_VTS_HAL_NEURALNETWORKS_H
 
-#include <android/hardware/neuralnetworks/1.2/IPreparedModel.h>
 #include <android/hardware/neuralnetworks/1.3/IDevice.h>
+#include <android/hardware/neuralnetworks/1.3/IPreparedModel.h>
 #include <android/hardware/neuralnetworks/1.3/types.h>
 #include <gtest/gtest.h>
 #include "1.0/Utils.h"
@@ -47,11 +47,10 @@
 // Create an IPreparedModel object. If the model cannot be prepared,
 // "preparedModel" will be nullptr instead.
 void createPreparedModel(const sp<IDevice>& device, const Model& model,
-                         sp<V1_2::IPreparedModel>* preparedModel);
+                         sp<IPreparedModel>* preparedModel);
 
 // Utility function to get PreparedModel from callback and downcast to V1_2.
-sp<V1_2::IPreparedModel> getPreparedModel_1_2(
-        const sp<implementation::PreparedModelCallback>& callback);
+sp<IPreparedModel> getPreparedModel_1_3(const sp<implementation::PreparedModelCallback>& callback);
 
 }  // namespace android::hardware::neuralnetworks::V1_3::vts::functional
 
diff --git a/neuralnetworks/1.3/vts/functional/include/1.3/Callbacks.h b/neuralnetworks/1.3/vts/functional/include/1.3/Callbacks.h
index 9376a92..fb19a84 100644
--- a/neuralnetworks/1.3/vts/functional/include/1.3/Callbacks.h
+++ b/neuralnetworks/1.3/vts/functional/include/1.3/Callbacks.h
@@ -137,7 +137,7 @@
      *     nullptr if the model was unable to be prepared.
      */
     Return<void> notify_1_3(V1_0::ErrorStatus status,
-                            const sp<V1_2::IPreparedModel>& preparedModel) override;
+                            const sp<V1_3::IPreparedModel>& preparedModel) override;
 
     /**
      * PreparedModelCallback::wait blocks until notify* has been called on the