Keep GPT4-Turbo for old vision agent

tests/test_llm.py

@@ -36,6 +36,19 @@ def test_chat_with_mock(openai_llm_mock): # noqa: F811
  )
 
 
+@pytest.mark.parametrize(
+ "openai_llm_mock_turbo", ["mocked response"], indirect=["openai_llm_mock_turbo"]
+)
+def openai_llm_mock_turbo(openai_llm_mock_2): # noqa: F811
+ llm = OpenAILLM()
+ response = llm.chat([{"role": "user", "content": "test prompt"}])
+ assert response == "mocked response"
+ openai_llm_mock_turbo.chat.completions.create.assert_called_once_with(
+ model="gpt-4-turbo",
+ messages=[{"role": "user", "content": "test prompt"}],
+ )
+
+
 @pytest.mark.parametrize(
  "openai_llm_mock", ["mocked response"], indirect=["openai_llm_mock"]
 )

vision_agent/agent/vision_agent.py

@@ -464,15 +464,17 @@ def __init__(
  report_progress_callback: a callback to report the progress of the agent. This is useful for streaming logs in a web application where multiple VisionAgent instances are running in parallel. This callback ensures that the progress are not mixed up.
  """
  self.task_model = (
- OpenAILLM(json_mode=True, temperature=0.1)
+ OpenAILLM(model_name="gpt-4-turbo", json_mode=True, temperature=0.0)
  if task_model is None
  else task_model
  )
  self.answer_model = (
- OpenAILLM(temperature=0.1) if answer_model is None else answer_model
+ OpenAILLM(model_name="gpt-4-turbo", temperature=0.0)
+ if answer_model is None
+ else answer_model
  )
  self.reflect_model = (
- OpenAILMM(json_mode=True, temperature=0.1)
+ OpenAILMM(model_name="gpt-4-turbo", json_mode=True, temperature=0.0)
  if reflect_model is None
  else reflect_model
  )