The Memory Layer for AI Agents - Drop-in memory infrastructure for AI agents and apps. Context that persists. Built for production.
Python agentic-memoryagentic-memory-systemagentsaiai-agentschatgptgenaillmlong-term-memorymemorymemory-managementpythonragstate-management
fix(llms): honor reasoning-model params in OpenAIStructuredLLM (#5458)
ly-wang1932575a65fc2b
diff --git a/mem0/llms/openai_structured.py b/mem0/llms/openai_structured.pyindex 6a9a14c36..2d5104039 100644--- a/mem0/llms/openai_structured.py+++ b/mem0/llms/openai_structured.py@@ -36,11 +36,8 @@ class OpenAIStructuredLLM(LLMBase):Returns:str: The generated response."""- params = {- "model": self.config.model,- "messages": messages,- "temperature": self.config.temperature,- }+ params = self._get_supported_params(messages=messages)+ params["model"] = self.config.modelif response_format:params["response_format"] = response_formatdiff --git a/tests/llms/test_openai_structured.py b/tests/llms/test_openai_structured.pynew file mode 100644index 000000000..4881d4b35--- /dev/null+++ b/tests/llms/test_openai_structured.py@@ -0,0 +1,51 @@+from unittest.mock import Mock, patch++import pytest++from mem0.configs.llms.openai import OpenAIConfig+from mem0.llms.openai_structured import OpenAIStructuredLLM+++def mock_openai_client():+ with patch("mem0.llms.openai_structured.OpenAI") as mock_openai:+ mock_client = Mock()+ mock_openai.return_value = mock_client+ yield mock_client+++def _mock_parse(mock_client, content="ok"):+ mock_response = Mock()+ mock_response.choices = [Mock(message=Mock(content=content))]+ mock_client.beta.chat.completions.parse.return_value = mock_response+++def test_reasoning_model_drops_temperature(mock_openai_client):+ """Reasoning models reject `temperature`; structured output must not send it."""+ config = OpenAIConfig(model="o3-mini", reasoning_effort="low")+ llm = OpenAIStructuredLLM(config)+ _mock_parse(mock_openai_client)++ llm.generate_response([{"role": "user", "content": "Hello"}])++ call_kwargs = mock_openai_client.beta.chat.completions.parse.call_args[1]+ assert "temperature" not in call_kwargs # reasoning models don't accept temperature+ assert "max_tokens" not in call_kwargs # also dropped for reasoning models+ assert "top_p" not in call_kwargs # also dropped for reasoning models+ assert call_kwargs["reasoning_effort"] == "low"+ assert call_kwargs["model"] == "o3-mini"+++def test_regular_model_sends_sampling_params(mock_openai_client):+ """Regular models still receive the standard sampling params."""+ config = OpenAIConfig(model="gpt-4o", temperature=0.3)+ llm = OpenAIStructuredLLM(config)+ _mock_parse(mock_openai_client)++ llm.generate_response([{"role": "user", "content": "Hello"}])++ call_kwargs = mock_openai_client.beta.chat.completions.parse.call_args[1]+ assert call_kwargs["temperature"] == 0.3+ assert "max_tokens" in call_kwargs # standard sampling params still forwarded+ assert "top_p" in call_kwargs+ assert call_kwargs["model"] == "gpt-4o"