diff --git a/tests/v1/entrypoints/openai/test_completion.py b/tests/v1/entrypoints/openai/test_completion.py
index ddab006d0..7faf25220 100644
--- a/tests/v1/entrypoints/openai/test_completion.py
+++ b/tests/v1/entrypoints/openai/test_completion.py
@@ -457,6 +457,18 @@ async def test_completion_stream_options(client: openai.AsyncOpenAI, model_name:
             )
             assert final_chunk.choices == []
 
+    # Test stream=True, stream_options={}
+    stream = await client.completions.create(
+        model=model_name,
+        prompt=prompt,
+        max_tokens=5,
+        temperature=0.0,
+        stream=True,
+        stream_options={},
+    )
+    async for chunk in stream:
+        assert chunk.usage is None
+
     # Test stream=False, stream_options=
     #     {"include_usage": None}
     with pytest.raises(BadRequestError):
diff --git a/vllm/entrypoints/openai/engine/protocol.py b/vllm/entrypoints/openai/engine/protocol.py
index ced89691f..02dad6c1f 100644
--- a/vllm/entrypoints/openai/engine/protocol.py
+++ b/vllm/entrypoints/openai/engine/protocol.py
@@ -159,7 +159,7 @@ AnyResponseFormat: TypeAlias = (
 
 
 class StreamOptions(OpenAIBaseModel):
-    include_usage: bool | None = True
+    include_usage: bool | None = False
     continuous_usage_stats: bool | None = False