diff --git a/benchmarks/experiment_runner.py b/benchmarks/experiment_runner.py
index d2e6c926525..7a7888f94b2 100644
--- a/benchmarks/experiment_runner.py
+++ b/benchmarks/experiment_runner.py
@@ -323,6 +323,9 @@ def timed_run(self, benchmark_experiment, benchmark_model):
     self._mark_step(benchmark_experiment)
     self._synchronize(benchmark_experiment)
 
+    # Clear XLA metrics before executing the model.
+    met.clear_metrics()
+
     enable_prof = self._args.profile_cuda
     metrics = OrderedDict()
     t_start = time.perf_counter()
@@ -363,9 +366,21 @@ def loop(prof=None):
     metrics["total_time"] = t_end - t_start
     metrics[
         "per_iter_time"] = metrics["total_time"] / self._args.iterations_per_run
+
     if benchmark_experiment.xla:
       metrics["trace_per_iter_time"] = t_trace / self._args.iterations_per_run
 
+      def ns_to_s(ns):
+        return ns * 1e-9
+
+      for m in ("CompileTime", "ExecuteTime"):
+        data = met.metric_data(m)
+        data = data if data is not None else (0, 0, [])
+        number, total_time, _ = data
+        # Time is measured in nano-seconds
+        metrics[f"xla_{m}_time_s"] = ns_to_s(total_time)
+        metrics[f"xla_{m}_number"] = number
+
     if enable_prof:
       self.collect_individual_ops(benchmark_experiment, metrics, prof)