deepseek-mcp-server

Overview

The DeepSeek MCP Server provides seamless integration with DeepSeek's advanced language models through the Model Context Protocol. It exposes key API surfaces including chat completions with thinking capabilities, function calling, FIM completion, model listing, and user balance checking. The server supports both local and remote deployment options with comprehensive documentation for various clients including Claude, Cursor, and Codex.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• chat_completion

  DeepSeek V4 chat with advanced features like thinking, reasoning effort, and conversation memory.

  Try:Write a Python function to calculate Fibonacci numbersTry:Explain quantum computing in simple termsTry:Help me draft a professional email to my client
• completion

  DeepSeek V4 Pro FIM (Fill in Middle) completion for code completion tasks.

  Try:Complete this Python function: def calculate_factorial(n):Try:Finish this JavaScript code: const array = [1, 2, 3];Try:Complete this SQL query: SELECT * FROM users WHERE
• list_models

  Read the live list of available DeepSeek models.

  Try:What models are available on DeepSeek?Try:List all available DeepSeek V4 modelsTry:Show me the available model options
• get_user_balance

  Read account balance and API availability status.

  Try:Check my DeepSeek account balanceTry:How much API quota do I have left?Try:Is my account active?
• reset_conversation

  Clear an in-memory conversation to start fresh.

  Try:Reset our current conversationTry:Start a new conversationTry:Clear the chat history
• list_conversations

  List all in-memory conversation IDs.

  Try:Show my current conversationsTry:List all conversation IDsTry:What chat sessions are active?

Comparable tools

Installation

DEEPSEEK_API_KEY="your-api-key" npx -y deepseek-mcp-server

docker pull docker.io/dmontgomery40/deepseek-mcp-server:0.5.0
docker run --rm -i -e DEEPSEEK_API_KEY="your-api-key" docker.io/dmontgomery40/deepseek-mcp-server:0.5.0

{
  "mcpServers": {
    "deepseek": {
      "command": "npx",
      "args": ["-y", "deepseek-mcp-server"],
      "env": {
        "DEEPSEEK_API_KEY": "your-api-key"
      }
    }
  }
}

codex mcp add deepseek --url https://deepseek-mcp.ragweld.com/mcp --bearer-token-env-var DEEPSEEK_MCP_AUTH_TOKEN

FAQ

What models does this server support?

The server supports DeepSeek V4 models including 'deepseek-v4-flash' and 'deepseek-v4-pro', with 'deepseek-v4-flash' as the default for chat completions.

Does this support function calling?

Yes, the chat completion tool supports function tools and can make function calls when appropriate.

Compare deepseek-mcp-server with