Checked other resources

• This is a bug, not a usage question. For questions, please use GitHub Discussions.
• I added a clear and detailed title that summarizes the issue.
• I read what a minimal reproducible example is (https://stackoverflow.com/help/minimal-reproducible-example).
• I included a self-contained, minimal example that demonstrates the issue INCLUDING all the relevant imports. The code run AS IS to reproduce the issue.

Example Code

llm = ChatOpenAI(
            api_key="xyz",  # Placeholder, AuthN headers will be used
            base_url=base_url,
            timeout=300,
            max_retries=3,
            seed=111
        )

agent = create_react_agent(
            model=llm_manager.get_llm(),
            tools=filtered_tools,
            debug=False,
            prompt=system_prompt,
            pre_model_hook=summarization_node,  # Use SummarizationNode
            checkpointer=checkpointer,
        )
        response = await agent.ainvoke(
            {"messages": [HumanMessage(content=prompt)]},
            config=config,
            extra_headers=proxy_header,
        )

Error Message and Stack Trace (if applicable)

Description

• I'm trying to pass additional args to agent which is used for all the llm calls, it don't work.
• It works, if I directly do llm.ainvoke(prompt, extra_headers=proxy_header)
• I have to create agent everytime now, for every request the agent receives.

System Info

langgraph = "0.3.31"
langchain-openai = "^0.3.4"
langchain-core = "^0.3.34"
langgraph-sdk = "^0.1.51"
jupyter = "^1.1.1"
langchain = "^0.3.18"
langfuse = "^2.59.2"
mcp = "^1.6.0"
langchain-mcp-adapters = "0.1.7"
langmem="0.0.27"