> ## Documentation Index
> Fetch the complete documentation index at: https://docs.langbot.app/llms.txt
> Use this file to discover all available pages before exploring further.

# DeepSeek R1 を WeChat・Discord・Telegram に5分で接続する方法

> A step-by-step guide to connecting DeepSeek R1/V3 API to WeChat (Personal/Enterprise), Discord, Telegram, and more using LangBot. No coding required.

この記事は LangBot Blog から同期されています。 [Blog の原文を読む](https://langbot.app/en/blog/connect-deepseek-to-wechat).

公開日：2026-02-21 · 著者：LangBot Team

> この記事の日本語版はまだ公開されていないため、英語版を掲載しています。

<img src="https://mintcdn.com/langbot/NbRS7zTwu8_8-9J1/images/articles/source/images/covers/deepseek-wechat.png?fit=max&auto=format&n=NbRS7zTwu8_8-9J1&q=85&s=cb4da1f6964fd2dcd2617055460246f2" alt="DeepSeek x WeChat Integration" width="1200" height="630" data-path="images/articles/source/images/covers/deepseek-wechat.png" />

DeepSeek has taken the AI world by storm. Its R1 reasoning model rivals OpenAI's o1 but is open-source and significantly cheaper.

However, the official DeepSeek app currently only supports "AI Search" in WeChat, and direct integration into group chats or private workflows is limited.

**What if you could have a fully functional DeepSeek R1 bot in your WeChat groups, Discord servers, and Telegram chats right now?**

With **LangBot**, you can. And it takes less than 5 minutes.

## Why LangBot?

LangBot is an open-source, production-grade IM bot platform. Unlike simple scripts or single-platform bots, LangBot gives you:

* **Unified Platform:** Connect once, deploy to WeChat, Discord, Telegram, Slack, Lark, and more.
* **Model Agnostic:** Use DeepSeek R1, V3, Claude 3.5, GPT-4, or local Ollama models.
* **No Coding:** robust WebUI for configuration.
* **Enterprise Features:** Knowledge base (RAG), Plugin system, and multi-user management.

## Prerequisites

1. **A Server or PC:** Docker installed (VPS, local computer, or Synology/NAS).
2. **DeepSeek API Key:**
   * **Official:** [platform.deepseek.com](https://platform.deepseek.com)
   * **SiliconFlow (Recommended for speed):** [cloud.siliconflow.cn](https://cloud.siliconflow.cn) (Faster R1 inference).

## Step 1: Deploy LangBot

If you haven't deployed LangBot yet, run this one command:

```bash theme={null}
git clone https://github.com/langbot-app/LangBot
cd LangBot/docker
docker compose up -d
```

Access your dashboard at `http://localhost:5300` (or your server IP).

## Step 2: Configure DeepSeek Model

1. Go to **Models** -> **Provider List**.
2. Find **DeepSeek** (or **SiliconFlow** / **OpenAI Compatible**).
3. Enter your API Key.
4. Click **Save**.

Now create a **Model Instance**:

* **Name:** `DeepSeek-R1-Bot`
* **Model:** `deepseek-reasoner` (for R1) or `deepseek-chat` (for V3).

## Step 3: Create a Chat Pipeline

LangBot uses "Pipelines" to manage bot logic.

1. Go to **Pipelines** -> **New Pipeline**.
2. Choose **Chat Pipeline**.
3. Select your `DeepSeek-R1-Bot` model.
4. (Optional) Add a System Prompt:
   > "You are a helpful assistant powered by DeepSeek R1. You think deeply before answering."

## Step 4: Connect to WeChat (or any platform)

### For WeChat (Personal / Wechaty)

*Note: Personal WeChat access relies on third-party libraries and carries some risk. For business use, we recommend Enterprise WeChat (WeCom).*

1. Go to **Bots** -> **New Bot**.
2. Select **WeChat**.
3. Choose the **GeweChat** or **Wechaty** adapter (depending on your preference).
4. Click **Save**.
5. Scan the QR code that appears in the logs or UI.

### For Discord / Telegram

1. Select **Discord** or **Telegram**.
2. Paste your **Bot Token** (from Discord Developer Portal or BotFather).
3. Click **Save**.

## Step 5: Test It Out!

Open your chat app and send a message. You should see DeepSeek R1 "thinking" (if supported) and replying with high-quality reasoning.

## Advanced: Using Search & Tools

DeepSeek is great, but DeepSeek with **Internet Access** is better.

In LangBot:

1. Go to **Plugins**.
2. Install **Tavily Search** or **Google Search**.
3. Add the plugin to your DeepSeek Pipeline.

Now your bot can search the web for real-time info *before* reasoning with DeepSeek R1!

***

**Ready to build?**
[Star us on GitHub](https://github.com/langbot-app/LangBot) or [Read the Docs](https://docs.langbot.app).