diff --git a/tests/e2e/online_serving/test_voxcpm2_tts_expansion.py b/tests/e2e/online_serving/test_voxcpm2_tts_expansion.py
index b889a9e57..d09570811 100644
--- a/tests/e2e/online_serving/test_voxcpm2_tts_expansion.py
+++ b/tests/e2e/online_serving/test_voxcpm2_tts_expansion.py
@@ -73,6 +73,7 @@ def test_voice_clone_streaming_001(omni_server, openai_client) -> None:
 
 @hardware_test(res={"cuda": "L4"}, num_cards=1)
 @pytest.mark.parametrize("omni_server", tts_server_params, indirect=True)
+@pytest.mark.skip(reason="#4335")
 def test_response_format_001(omni_server, openai_client) -> None:
     """
     Test voice-clone non-stream PCM output via OpenAI API.