diff --git a/torch/csrc/distributed/c10d/init.cpp b/torch/csrc/distributed/c10d/init.cpp
index 6f5ae85ca9a6..d6fb2bd047a0 100644
--- a/torch/csrc/distributed/c10d/init.cpp
+++ b/torch/csrc/distributed/c10d/init.cpp
@@ -3369,6 +3369,24 @@ Example::
   auto processGroupXCCL =
       intrusive_ptr_no_gil_destructor_class_<::c10d::ProcessGroupXCCL>(
           module, "ProcessGroupXCCL", backend)
+          .def(
+              py::init([](const c10::intrusive_ptr<::c10d::Store>& store,
+                          int rank,
+                          int size,
+                          c10::intrusive_ptr<::c10d::ProcessGroupXCCL::Options>
+                              options) {
+                // gil_scoped_release is not safe as a call_guard in init.
+                // https://github.com/pybind/pybind11/issues/5473
+                py::gil_scoped_release nogil{};
+
+                return c10::make_intrusive<::c10d::ProcessGroupXCCL>(
+                    store, rank, size, std::move(options));
+              }),
+              py::arg("store"),
+              py::arg("rank"),
+              py::arg("size"),
+              py::arg("options"),
+              R"(Create a new ProcessGroupXCCL instance.)")
           .def(
               py::init([](const c10::intrusive_ptr<::c10d::Store>& store,
                           int rank,
@@ -3377,12 +3395,50 @@ Example::
                 // https://github.com/pybind/pybind11/issues/5473
                 py::gil_scoped_release nogil{};
 
+                auto options = ::c10d::ProcessGroupXCCL::Options::create();
+                options->is_high_priority_stream = false;
                 return c10::make_intrusive<::c10d::ProcessGroupXCCL>(
-                    store, rank, size);
+                    store, rank, size, options);
               }),
               py::arg("store"),
               py::arg("rank"),
-              py::arg("size"));
+              py::arg("size"),
+              R"(Create a new ProcessGroupXCCL instance.)")
+          .def(
+              "comm_split_count",
+              &::c10d::ProcessGroupXCCL::getCommSplitCounter)
+          .def_property_readonly(
+              "options",
+              &::c10d::ProcessGroupXCCL::getOptions,
+              R"(Return the options used to create this ProcessGroupXCCL instance.)")
+          .def_property(
+              "bound_device_id",
+              &::c10d::ProcessGroupXCCL::getBoundDeviceId,
+              &::c10d::ProcessGroupXCCL::setBoundDeviceId,
+              R"(Return the bound device id.)")
+          .def(
+              "perform_nocolor_split",
+              &::c10d::ProcessGroupXCCL::performNocolorSplit)
+          .def(
+              "_is_initialized",
+              &::c10d::ProcessGroupXCCL::isInitialized,
+              py::call_guard<py::gil_scoped_release>());
+  intrusive_ptr_class_<::c10d::ProcessGroupXCCL::Options>(
+      processGroupXCCL, "Options", backendOptions)
+      .def(py::init<bool>(), py::arg("is_high_priority_stream") = false)
+      .def_readwrite("config", &::c10d::ProcessGroupXCCL::Options::config)
+      .def_readwrite(
+          "is_high_priority_stream",
+          &::c10d::ProcessGroupXCCL::Options::is_high_priority_stream)
+      .def_readwrite(
+          "split_from", &::c10d::ProcessGroupXCCL::Options::split_from)
+      .def_readwrite(
+          "split_color", &::c10d::ProcessGroupXCCL::Options::split_color)
+      .def_readwrite(
+          "global_ranks_in_group",
+          &::c10d::ProcessGroupXCCL::Options::global_ranks_in_group)
+      .def_readwrite(
+          "group_name", &::c10d::ProcessGroupXCCL::Options::group_name);
 #endif
 
 #ifdef USE_C10D_UCC
diff --git a/torch/distributed/distributed_c10d.py b/torch/distributed/distributed_c10d.py
index 866658515a74..bfbf1ee73436 100644
--- a/torch/distributed/distributed_c10d.py
+++ b/torch/distributed/distributed_c10d.py
@@ -2033,8 +2033,18 @@ def _new_process_group_helper(
         elif backend_str == Backend.XCCL:
             if not is_xccl_available():
                 raise RuntimeError("Distributed package doesn't have XCCL built in")
+            if backend_options is not None:
+                assert isinstance(backend_options, ProcessGroupXCCL.Options), (
+                    "Expected backend_options argument to be of type ProcessGroupXCCL.Options"
+                )
+            else:
+                # default backend_options for XCCL
+                backend_options = ProcessGroupXCCL.Options()
+                backend_options.is_high_priority_stream = False
+            backend_options.global_ranks_in_group = global_ranks_in_group
+            backend_options.group_name = group_name
             backend_class = ProcessGroupXCCL(
-                backend_prefix_store, group_rank, group_size
+                backend_prefix_store, group_rank, group_size, backend_options
             )
             backend_type = ProcessGroup.BackendType.XCCL
         else:
@@ -5042,7 +5052,7 @@ def split_group(
         )
 
     parent_group_rank = parent_global_to_group_ranks[global_rank]
-    parent_backend = parent_pg._get_backend(torch.device("cuda"))
+    parent_backend = parent_pg._get_backend(device_id)
 
     # if the parent backend does not support splitting, raise error
     # currently this API only support NCCL backend
@@ -5123,6 +5133,15 @@ def split_group(
         backend_class = ProcessGroupNCCL(
             prefix_store, group_rank, len(my_group), pg_options
         )
+    elif parent_backend_str == Backend.XCCL:
+        backend_type = ProcessGroup.BackendType.XCCL
+        if not isinstance(pg_options, ProcessGroupXCCL.Options):
+            raise RuntimeError(
+                "Expected pg_options argument to be of type ProcessGroupXCCL.Options"
+            )
+        backend_class = ProcessGroupXCCL(
+            prefix_store, group_rank, len(my_group), pg_options
+        )
     else:
         assert parent_backend_str.upper() in Backend._plugins, (
             f"Unknown c10d backend type {parent_backend_str.upper()}"
@@ -5143,7 +5162,9 @@ def split_group(
     pg._set_default_backend(backend_type)
     backend_class._set_sequence_number_for_group()
 
-    pg._register_backend(torch.device("cuda"), backend_type, backend_class)
+    pg._register_backend(
+        torch.accelerator.current_accelerator(), backend_type, backend_class
+    )
 
     # set group_name and group_desc to backend
     assert group_name is not None