ChatDeepSeek

这将帮助你开始使用 DeepSeek 聊天模型。有关 ChatDeepSeek 所有功能和配置的详细文档，请参阅 API 参考。

¥This will help you getting started with DeepSeek chat models. For detailed documentation of all ChatDeepSeek features and configurations head to the API reference.

概述

¥Overview

集成详情

¥Integration details

Class	Package	Local	Serializable	PY support	Package downloads	Package latest
ChatDeepSeek	@langchain/deepseek	❌ (see Ollama)	beta	✅

模型特性

¥Model features

有关如何使用特定功能的指南，请参阅下方表格标题中的链接。

¥See the links in the table headers below for guides on how to use specific features.

Tool calling	Structured output	JSON mode	Image input	Audio input	Video input	Token-level streaming	Token usage	Logprobs
✅	✅	✅	❌	❌	❌	✅	✅	✅

请注意，自 2025 年 1 月 27 日起，deepseek-reasoner 目前不支持工具调用和结构化输出。

¥Note that as of 1/27/25, tool calling and structured output are not currently supported for deepseek-reasoner.

设置

¥Setup

要访问 DeepSeek 模型，你需要创建一个 DeepSeek 账户、获取 API 密钥并安装 @langchain/deepseek 集成包。

¥To access DeepSeek models you’ll need to create a DeepSeek account, get an API key, and install the @langchain/deepseek integration package.

你还可以通过 Together AI 或 Ollama 等提供程序访问 DeepSeek API。

¥You can also access the DeepSeek API through providers like Together AI or Ollama.

凭证

¥Credentials

前往 https://deepseek.com/ 注册 DeepSeek 并生成 API 密钥。完成此操作后，请设置 DEEPSEEK_API_KEY 环境变量：

¥Head to https://deepseek.com/ to sign up to DeepSeek and generate an API key. Once you’ve done this set the DEEPSEEK_API_KEY environment variable:

export DEEPSEEK_API_KEY="your-api-key"

如果你想自动追踪模型调用，也可以通过取消注释以下内容来设置你的 LangSmith API 密钥：

¥If you want to get automated tracing of your model calls you can also set your LangSmith API key by uncommenting below:

# export LANGSMITH_TRACING="true"
# export LANGSMITH_API_KEY="your-api-key"

安装

¥Installation

LangChain ChatDeepSeek 集成位于 @langchain/deepseek 包中：

¥The LangChain ChatDeepSeek integration lives in the @langchain/deepseek package:

tip

请参阅 此部分包含有关安装集成包的一般说明。

¥See this section for general instructions on installing integration packages.

npm

npm i @langchain/deepseek @langchain/core

yarn

yarn add @langchain/deepseek @langchain/core

pnpm

pnpm add @langchain/deepseek @langchain/core

实例化

¥Instantiation

现在我们可以实例化我们的模型对象并生成聊天补全：

¥Now we can instantiate our model object and generate chat completions:

import { ChatDeepSeek } from "@langchain/deepseek";

const llm = new ChatDeepSeek({
  model: "deepseek-reasoner",
  temperature: 0,
  // other params...
});

const aiMsg = await llm.invoke([
  [
    "system",
    "You are a helpful assistant that translates English to French. Translate the user sentence.",
  ],
  ["human", "I love programming."],
]);
aiMsg;

AIMessage {
  "id": "e2874482-68a7-4552-8154-b6a245bab429",
  "content": "J'adore la programmation.",
  "additional_kwargs": {,
    "reasoning_content": "...",
  },
  "response_metadata": {
    "tokenUsage": {
      "promptTokens": 23,
      "completionTokens": 7,
      "totalTokens": 30
    },
    "finish_reason": "stop",
    "model_name": "deepseek-reasoner",
    "usage": {
      "prompt_tokens": 23,
      "completion_tokens": 7,
      "total_tokens": 30,
      "prompt_tokens_details": {
        "cached_tokens": 0
      },
      "prompt_cache_hit_tokens": 0,
      "prompt_cache_miss_tokens": 23
    },
    "system_fingerprint": "fp_3a5770e1b4"
  },
  "tool_calls": [],
  "invalid_tool_calls": [],
  "usage_metadata": {
    "output_tokens": 7,
    "input_tokens": 23,
    "total_tokens": 30,
    "input_token_details": {
      "cache_read": 0
    },
    "output_token_details": {}
  }
}

console.log(aiMsg.content);

J'adore la programmation.

链接

¥Chaining

我们可以使用如下提示模板对我们的模型进行 chain 操作：

¥We can chain our model with a prompt template like so:

import { ChatPromptTemplate } from "@langchain/core/prompts";

const prompt = ChatPromptTemplate.fromMessages([
  [
    "system",
    "You are a helpful assistant that translates {input_language} to {output_language}.",
  ],
  ["human", "{input}"],
]);

const chain = prompt.pipe(llm);
await chain.invoke({
  input_language: "English",
  output_language: "German",
  input: "I love programming.",
});

AIMessage {
  "id": "6e7f6f8c-8d7a-4dad-be07-425384038fd4",
  "content": "Ich liebe es zu programmieren.",
  "additional_kwargs": {,
    "reasoning_content": "...",
  },
  "response_metadata": {
    "tokenUsage": {
      "promptTokens": 18,
      "completionTokens": 9,
      "totalTokens": 27
    },
    "finish_reason": "stop",
    "model_name": "deepseek-reasoner",
    "usage": {
      "prompt_tokens": 18,
      "completion_tokens": 9,
      "total_tokens": 27,
      "prompt_tokens_details": {
        "cached_tokens": 0
      },
      "prompt_cache_hit_tokens": 0,
      "prompt_cache_miss_tokens": 18
    },
    "system_fingerprint": "fp_3a5770e1b4"
  },
  "tool_calls": [],
  "invalid_tool_calls": [],
  "usage_metadata": {
    "output_tokens": 9,
    "input_tokens": 18,
    "total_tokens": 27,
    "input_token_details": {
      "cache_read": 0
    },
    "output_token_details": {}
  }
}

API 参考

¥API reference

有关 ChatDeepSeek 所有功能和配置的详细文档，请参阅 API 参考：https://api.js.langchain.com/classes/_langchain_deepseek.ChatDeepSeek.html

¥For detailed documentation of all ChatDeepSeek features and configurations head to the API reference: https://api.js.langchain.com/classes/_langchain_deepseek.ChatDeepSeek.html