ChatDeepSeek

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

Overview

Integration details

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

Model features

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
✅	✅	✅	❌	❌	❌	✅	✅	✅

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

Setup

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

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

Credentials

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"

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"

Installation

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

tip

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

Instantiation

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.

Chaining

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 reference

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

Related

• Chat model conceptual guide
• Chat model how-to guides