From 1c910875d2311473c2dfd84257ab61b5639ae8ea Mon Sep 17 00:00:00 2001
From: Taylor Howell <taylorhowell@google.com>
Date: Mon, 24 Nov 2025 09:12:38 -0800
Subject: [PATCH] MuJoCo Warp documentation: Differentiability.

PiperOrigin-RevId: 836250165
Change-Id: I40c21674c896ed4108f17192f16aaaf2664f03be
---
 doc/mjwarp/index.rst | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/doc/mjwarp/index.rst b/doc/mjwarp/index.rst
index d2bc1e92..dbf68489 100644
--- a/doc/mjwarp/index.rst
+++ b/doc/mjwarp/index.rst
@@ -373,6 +373,16 @@ For examples that train policies with MJWarp physics, please see:
 - `mjlab <https://github.com/mujocolab/mjlab>`__: Train directly with MJWarp using PyTorch.
 - `MuJoCo Playground <https://github.com/google-deepmind/mujoco_playground>`__: Train via :ref:`MJX API <mjx>`.
 
+Features
+--------
+
+**Is MJWarp differentiable?**
+
+No. MJWarp is not currently differentiable via
+Warp's `automatic differentiation <https://nvidia.github.io/warp/modules/differentiability.html#differentiability>`__
+functionality. Updates from the team related to enabling automatic differentiation for MJWarp are tracked in this
+`GitHub issue <https://github.com/google-deepmind/mujoco_warp/issues/500>`__.
+
 Orientation representation
 --------------------------