Configure DeepSeek in Cursor (Official / OpenRouter / SiliconFlow)

DeepSeek V3/R1 can be easily integrated into Cursor IDE to enhance your coding experience with top-tier AI capabilities at a lower cost.

This guide covers how to set up DeepSeek in Cursor using the official API or third-party providers like OpenRouter and SiliconFlow.

Quick Parameters

Note: Cursor integrates DeepSeek via its "OpenAI Compatible" interface.

Configuration Steps

1. Open Settings: Click the gear icon in the top right, or press Cmd + , (Mac) / Ctrl + , (Win) to open Cursor Settings.
2. Go to Models: Select Models from the sidebar menu.
3. Add Custom Model:
   • Under "Model Names", type the model name (e.g., deepseek-chat) and click the + button to add and enable it.
4. Configure API Key & Base URL:
   • Scroll down to the "OpenAI API Key" section.
   • Crucial Step: Click "Override OpenAI Base URL".
   • Enter the Base URL: https://api.deepseek.com (or your provider's URL).
   • Enter your API Key (sk-xxxx).
5. Verify: Click the "Verify" button. A green checkmark indicates success!

Configuration Examples

1. DeepSeek Official (Recommended)

Most stable and cost-effective.

{
  "baseUrl": "https://api.deepseek.com",
  "apiKey": "sk-your-deepseek-key",
  "model": "deepseek-chat"
}

2. OpenRouter

Great for accessing multiple models.

{
  "baseUrl": "https://openrouter.ai/api/v1",
  "apiKey": "sk-your-openrouter-key",
  "model": "deepseek/deepseek-chat"
}

FAQ

Q: "Connection failed" error? A: Please check:

1. Is the Base URL correct? (Must include https://).
2. Do not use a trailing slash if not needed (standard is usually without, but some tools differ. Cursor handles both well).
3. Do you have sufficient balance in your API account?

Q: How to use DeepSeek R1 (Reasoning)? A: Simply add deepseek-reasoner to your model list in Cursor, then select it in the chat interface (Cmd+K / Cmd+L).

Q: Is deepseek-chat V2 or V3? A: The official deepseek-chat endpoint now points to the latest DeepSeek V3 model.