diff --git a/tests/entrypoints/multimodal/openai/chat_completion/test_audio.py b/tests/entrypoints/multimodal/openai/chat_completion/test_audio.py
index a1c13d9e33..d806100e8e 100644
--- a/tests/entrypoints/multimodal/openai/chat_completion/test_audio.py
+++ b/tests/entrypoints/multimodal/openai/chat_completion/test_audio.py
@@ -10,7 +10,6 @@ import pytest_asyncio
 from tests.utils import RemoteOpenAIServer
 from vllm.assets.audio import AudioAsset
 from vllm.multimodal.utils import encode_audio_base64, encode_audio_url, fetch_audio
-from vllm.platforms import current_platform
 
 MODEL_NAME = "fixie-ai/ultravox-v0_5-llama-3_2-1b"
 TEST_AUDIO_URLS = [
@@ -19,10 +18,6 @@ TEST_AUDIO_URLS = [
 ]
 MAXIMUM_AUDIOS = 2
 
-# Disable prefix caching on ROCm to reduce non-determinism in
-# streaming-vs-non-streaming comparisons.
-_ROCM_ARGS = ["--no-enable-prefix-caching"] if current_platform.is_rocm() else []
-
 
 @pytest.fixture(scope="module")
 def server():
@@ -37,7 +32,7 @@ def server():
         "--trust-remote-code",
         "--limit-mm-per-prompt",
         json.dumps({"audio": MAXIMUM_AUDIOS}),
-        *_ROCM_ARGS,
+        "--no-enable-prefix-caching",
     ]
 
     with RemoteOpenAIServer(MODEL_NAME, args) as remote_server: