diff --git a/doc/source/ray-core/configure.rst b/doc/source/ray-core/configure.rst
index d204ddb6fb78..46c8d38f9421 100644
--- a/doc/source/ray-core/configure.rst
+++ b/doc/source/ray-core/configure.rst
@@ -12,6 +12,8 @@ and from the command line. Take a look at the ``ray.init`` `documentation
 .. important:: For the multi-node setting, you must first run ``ray start`` on the command line to start the Ray cluster services on the machine before ``ray.init`` in Python to connect to the cluster services. On a single machine, you can run ``ray.init()`` without ``ray start``, which will both start the Ray cluster services and connect to them.
 
 
+.. _cluster-resources:
+
 Cluster Resources
 -----------------
 
diff --git a/doc/source/ray-core/miscellaneous.rst b/doc/source/ray-core/miscellaneous.rst
index c75777ea86e9..9c20ce31282c 100644
--- a/doc/source/ray-core/miscellaneous.rst
+++ b/doc/source/ray-core/miscellaneous.rst
@@ -75,6 +75,8 @@ appear as the task name in the logs.
 .. image:: images/task_name_dashboard.png
 
 
+.. _accelerator-types:
+
 Accelerator Types
 ------------------
 
diff --git a/doc/source/serve/scaling-and-resource-allocation.md b/doc/source/serve/scaling-and-resource-allocation.md
index a7d6b79d444f..47148b8213aa 100644
--- a/doc/source/serve/scaling-and-resource-allocation.md
+++ b/doc/source/serve/scaling-and-resource-allocation.md
@@ -123,6 +123,29 @@ def func_2(*args):
 
 In this example, each replica of each deployment will be allocated 0.5 GPUs.  The same can be done to multiplex over CPUs, using `"num_cpus"`.
 
+### Custom Resources, Accelerator types, and more
+
+You can also specify {ref}`custom resources <cluster-resources>` in `ray_actor_options`, for example to ensure that a deployment is scheduled on a specific node.
+For example, if you have a deployment that requires 2 units of the `"custom_resource"` resource, you can specify it like this:
+
+```python
+@serve.deployment(ray_actor_options={"resources": {"custom_resource": 2}})
+def func(*args):
+    return do_something_with_my_custom_resource()
+```
+
+You can also specify {ref}`accelerator types <accelerator-types>` via the `accelerator_type` parameter in `ray_actor_options`.
+
+Below is the full list of supported options in `ray_actor_options`; please see the relevant Ray Core documentation for more details about each option:
+
+- `accelerator_type`
+- `memory`
+- `num_cpus`
+- `num_gpus`
+- `object_store_memory`
+- `resources`
+- `runtime_env`
+
 (serve-omp-num-threads)=
 
 ## Configuring Parallelism with OMP_NUM_THREADS
diff --git a/python/ray/serve/api.py b/python/ray/serve/api.py
index 611b665a408c..7a2754954c93 100644
--- a/python/ray/serve/api.py
+++ b/python/ray/serve/api.py
@@ -317,7 +317,9 @@ def deployment(
             a '/' unless they're the root (just '/'), which acts as a
             catch-all.
         ray_actor_options: Options to be passed to the Ray actor
-            constructor such as resource requirements.
+            constructor such as resource requirements. Valid options are
+            `accelerator_type`, `memory`, `num_cpus`, `num_gpus`,
+            `object_store_memory`, `resources`, and `runtime_env`.
         user_config (Optional[Any]): Config to pass to the
             reconfigure method of the deployment. This can be updated
             dynamically without changing the version of the deployment and
diff --git a/python/ray/serve/config.py b/python/ray/serve/config.py
index c11d233d17d7..34187b99e338 100644
--- a/python/ray/serve/config.py
+++ b/python/ray/serve/config.py
@@ -390,7 +390,8 @@ def _validate_ray_actor_options(self) -> None:
                 f'Got invalid type "{type(self.ray_actor_options)}" for '
                 "ray_actor_options. Expected a dictionary."
             )
-
+        # Please keep this in sync with the docstring for the ray_actor_options
+        # kwarg in api.py.
         allowed_ray_actor_options = {
             # Resource options
             "accelerator_type",