Checkboxes

• 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.
• This is related to langchain-openai package.
• I posted a self-contained, minimal, reproducible example.

Package

• langchain-openai

Related Issues

• #35006
• #34938

Reproduction Steps

import { ChatOpenAI } from "@langchain/openai";

const llm = new ChatOpenAI({
  model: "deepseek-reasoner",
  apiKey: process.env.DEEPSEEK_API_KEY,
  baseURL: "https://api.deepseek.com"
});

const stream = await llm.stream("What is 2+2?");

for await (const chunk of stream) {
  // chunk.additional_kwargs.reasoning_content is always undefined
  console.log("reasoning:", chunk.additional_kwargs?.reasoning_content);
}

Expected: reasoning_content should contain the model's thinking process. Actual: reasoning_content is always undefined.

Description

LangChain's ChatOpenAI doesn't extract reasoning_content from DeepSeek's streaming responses. The reasoning_content field is used by DeepSeek models (like deepseek-reasoner) to stream their thinking process.

Suggested Fix

In libs/langchain-openai/src/chat_models/completions.ts, function convertCompletionsDeltaToBaseMessageChunk, add:

// Extract reasoning_content (DeepSeek specific)
if (delta.reasoning_content) {
  additional_kwargs.reasoning_content = delta.reasoning_content;
}

System Info

• langchain-openai: ^0.1.0
• Node.js: 18+
• DeepSeek API

PR

Fix submitted: #35520