fix(openai): ainvoke uses async _aget_response; add async tests

libs/langchain/langchain/agents/openai_assistant/base.py

@@ -360,6 +360,7 @@ def invoke(
             run_manager.on_chain_error(e)
             raise
         try:
+            # Use sync response handler in sync invoke
             response = self._get_response(run)
         except BaseException as e:
             run_manager.on_chain_error(e, metadata=run.dict())
@@ -496,7 +497,8 @@ async def ainvoke(
             run_manager.on_chain_error(e)
             raise
         try:
-            response = self._get_response(run)
+            # Use async response handler in async ainvoke
+            response = await self._aget_response(run)
         except BaseException as e:
             run_manager.on_chain_error(e, metadata=run.dict())
             raise

libs/langchain/tests/unit_tests/agents/test_openai_assistant.py

@@ -17,7 +17,7 @@ def _create_mock_client(*_: Any, use_async: bool = False, **__: Any) -> Any:
 
 @pytest.mark.requires("openai")
 def test_user_supplied_client() -> None:
-    import openai
+    openai = pytest.importorskip("openai")
 
     client = openai.AzureOpenAI(
         azure_endpoint="azure_endpoint",
@@ -48,6 +48,82 @@ def test_create_assistant() -> None:
     assert isinstance(assistant, OpenAIAssistantRunnable)
 
 
+@pytest.mark.requires("openai")
+@patch(
+    "langchain.agents.openai_assistant.base._get_openai_async_client",
+    new=partial(_create_mock_client, use_async=True),
+)
+async def test_ainvoke_uses_async_response_completed() -> None:
+    # Arrange a runner with mocked async client and a completed run
+    assistant = OpenAIAssistantRunnable(
+        assistant_id="assistant_id",
+        client=_create_mock_client(),
+        async_client=_create_mock_client(use_async=True),
+        as_agent=False,
+    )
+    mock_run = MagicMock()
+    mock_run.id = "run-id"
+    mock_run.thread_id = "thread-id"
+    mock_run.status = "completed"
+
+    # await_for_run returns a completed run
+    await_for_run_mock = AsyncMock(return_value=mock_run)
+    # async messages list returns messages belonging to run
+    msg = MagicMock()
+    msg.run_id = "run-id"
+    msg.content = []
+    list_mock = AsyncMock(return_value=[msg])
+
+    with patch.object(assistant, "_await_for_run", await_for_run_mock), patch.object(
+        assistant.async_client.beta.threads.messages,
+        "list",
+        list_mock,
+    ):
+        # Act
+        result = await assistant.ainvoke({"content": "hi"})
+
+    # Assert: returns messages list (non-agent path) and did not block
+    assert isinstance(result, list)
+    list_mock.assert_awaited()
+
+
+@pytest.mark.requires("openai")
+@patch(
+    "langchain.agents.openai_assistant.base._get_openai_async_client",
+    new=partial(_create_mock_client, use_async=True),
+)
+async def test_ainvoke_uses_async_response_requires_action_agent() -> None:
+    # Arrange a runner with mocked async client and requires_action run
+    assistant = OpenAIAssistantRunnable(
+        assistant_id="assistant_id",
+        client=_create_mock_client(),
+        async_client=_create_mock_client(use_async=True),
+        as_agent=True,
+    )
+    mock_run = MagicMock()
+    mock_run.id = "run-id"
+    mock_run.thread_id = "thread-id"
+    mock_run.status = "requires_action"
+
+    # Fake tool call structure
+    tool_call = MagicMock()
+    tool_call.id = "tool-id"
+    tool_call.function.name = "foo"
+    tool_call.function.arguments = '{\n  "x": 1\n}'
+    mock_run.required_action.submit_tool_outputs.tool_calls = [tool_call]
+
+    await_for_run_mock = AsyncMock(return_value=mock_run)
+
+    # Act
+    with patch.object(assistant, "_await_for_run", await_for_run_mock):
+        result = await assistant.ainvoke({"content": "hi"})
+
+    # Assert: returns list of OpenAIAssistantAction
+    assert isinstance(result, list)
+    assert result
+    assert getattr(result[0], "tool", None) == "foo"
+
+
 @pytest.mark.requires("openai")
 @patch(
     "langchain.agents.openai_assistant.base._get_openai_async_client",