Submission checklist

• This is a bug, not a usage question.
• I added a clear and descriptive title that summarizes this issue.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).
• This is not related to the langchain-community package.
• I posted a self-contained, minimal, reproducible example. A maintainer can copy it and run it AS IS.

Package (Required)

• langchain
• langchain-openai
• langchain-anthropic
• langchain-classic
• langchain-core
• langchain-model-profiles
• langchain-tests
• langchain-text-splitters
• langchain-chroma
• langchain-deepseek
• langchain-exa
• langchain-fireworks
• langchain-groq
• langchain-huggingface
• langchain-mistralai
• langchain-nomic
• langchain-ollama
• langchain-openrouter
• langchain-perplexity
• langchain-qdrant
• langchain-xai
• Other / not sure / general

Related Issues / PRs

No response

Reproduction Steps / Example Code (Python)

from langchain.agents import create_agent
  from langchain_core.tools import tool
  from langchain.chat_models import init_chat_model

  @tool(description="query weather")
  def get_weather() -> str:
      return "sunny"

  model = init_chat_model(model="deepseek-v4-flash")

  agent = create_agent(
      model=model,
      tools=[get_weather],
      system_prompt="You are a helpful assistant.",
  )

  res = agent.invoke({
      "messages": [
          {"role": "user", "content": "What is the weather tomorrow?"},
      ]
  })

  for msg in res["messages"]:
      print(type(msg).__name__, msg.content)

Error Message and Stack Trace (if applicable)

openai.BadRequestError: Error code: 400
  {'error': {'message': 'The `reasoning_content` in the thinking mode must be passed back to the API.', 'type': 'invalid_request_error',
  'param': None, 'code': 'invalid_request_error'}}

  Traceback (most recent call last):
    File "test_agent.py", line 17, in <module>
      res = agent.invoke({"messages": [{"role": "user", "content": "..."}]})
    File ".../langgraph/pregel/main.py", line 3365, in invoke
      for chunk in self.stream(...)
    ...
    File ".../langchain_deepseek/chat_models.py", line 356, in _generate
      return super()._generate(...)
    File ".../langchain_openai/chat_models/base.py", line 1653, in _generate
      _handle_openai_bad_request(e)
  openai.BadRequestError: Error code: 400 - {'error': {'message': 'The `reasoning_content` in the thinking mode must be passed back to the
  API.'}}

Description

用 langchain-deepseek（v1.0.1）调用deepseek-v4-flash时，跑 agent.invoke() 会在第二次内部调用 LLM 的时候挂掉，报 400 错误。 原因：DeepSeek API 要求，如果你在后续请求里要传回上一条助手消息，那这条消息里的 reasoning_content 字段必须原样带上。但是 ChatDeepSeek._get_request_payload() 用的是父类 BaseChatOpenAI._get_request_payload()，它调用的 _convert_message_to_dict() 在把 AIMessage 转成字典的时候，会把 additional_kwargs（包括 reasoning_content）给扔掉。 ChatDeepSeek._create_chat_result() 其实是能从 API 返回的结果里拿到 reasoning_content，并正确存放到 AIMessage.additional_kwargs["reasoning_content"] 里面的。但问题是，后面再发请求的时候，_get_request_payload() 里并没有对应的逻辑把它再塞回去。 修复方法：在 _get_request_payload() 里打了个补丁，让它先去拿到原始的 AIMessage 对象，然后从 additional_kwargs 里把 reasoning_content 捞出来，在发请求前塞到 API 的载荷字典里。

System Info

Package: langchain-deepseek 1.0.1 Model: deepseek-v4-flash (also affects deepseek-r1) OS: Windows 11 Python: 3.10

我是一名来自中国天津的大学生，目前正在学习LangChain框架。我在构建一个简单用于测试目的agent时发现了此问题。