Transforming Programming Efficiency with DeepSeek R1 in VSCode

Hello everyone, I am a rural programmer, an independent developer, and an industry observer, Chen Suiyi, the “Tiger of Frontend”.

  • • Personal website 1️⃣: https://chensuiyi.me
  • • Personal website 2️⃣: https://me.yicode.tech

If this article inspires or helps you, feel free to support me with a like, comment, or share. Thank you!

Transforming Programming Efficiency with DeepSeek R1 in VSCode
DeepSeek-R1

It’s incredible, everyone. DeepSeek has just released <span>DeepSeek-R1</span>, which performs on par with <span>OpenAI o1 official version</span>.

Transforming Programming Efficiency with DeepSeek R1 in VSCode
Data Comparison

You can see that its capabilities are outstanding.

Transforming Programming Efficiency with DeepSeek R1 in VSCode
Pricing

<span>DeepSeek-r1</span> API pricing:

  • • Per million input tokens <span>1 Yuan</span> (cache hit)
  • • Per million input tokens <span>4 Yuan</span> (cache miss)
  • • Per million output tokens <span>16 Yuan</span>.

<span>OpenAI-o1</span> API pricing:

  • • Per million input tokens $15 (<span>110 Yuan</span>) (cache hit)
  • • Per million input tokens $7.5 (<span>55 Yuan</span>) (cache miss)
  • • Per million output tokens $60 (<span>438 Yuan</span>)

In terms of pricing, comparing to <span>o1</span> is a huge bargain.

Transforming Programming Efficiency with DeepSeek R1 in VSCode
Bar Chart

If it’s still not intuitive, take a look at the comparison above, it’s clear at a glance.

Some netizens commented:

This is the real OpenAI.

It does make some sense.

As a programmer, using such a cost-effective model for enhancing my coding efficiency is wonderful.

Transforming Programming Efficiency with DeepSeek R1 in VSCode
VSCode Update Log

I am a fan of <span>VSCode</span>, I share the update content every month.

So, if I could use <span>DeepSeek r1</span> in <span>VSCode</span>, that would be perfect.

Someone recommended I use <span>Cursor</span> and <span>Windsurf</span>. Honestly, although they are both derived from <span>VSCode</span>, I still prefer the original <span>VSCode</span>.

Transforming Programming Efficiency with DeepSeek R1 in VSCode
VSCode Extension Roo Cline

After some searching, I finally found a magical tool called <span>Roo Cline</span>.

This is a VSCode extension that connects to different AI large models to provide suggestions and modifications for our programming.

Speaking of <span>Roo Cline</span>, I must mention <span>Cline</span>.

Transforming Programming Efficiency with DeepSeek R1 in VSCode
Cline

I have previously shared about <span>cline</span>, and you can find more of my past articles at the end of this article.

<span>Roo Cline</span> is a fork of <span>Cline</span>, based on its open-source code, starting a new project maintained by a different group of people.

Compared to <span>Cline</span>, <span>Roo Cline</span> has the following features:

Function Enhancements and Improvements

  • <span>More aggressive update cycle</span>: <span>Roo Cline</span> has a higher update frequency, supporting more experimental features such as image dragging, message deletion, feedback sounds, browser calls, etc.
  • <span>Multi-language support</span>: <span>Roo Cline</span> supports multiple language interfaces (like Chinese, English, Japanese, Spanish, etc.), making it convenient for global developers to use.
  • <span>Richer model support</span>: <span>Roo Cline</span> not only supports the models originally available in Cline (such as <span>Claude 3.5 Sonnet</span>), but also adds support for <span>DeepSeek</span>, <span>AWS Bedrock</span>, <span>Google Gemini</span>, etc., and can quickly follow up on new model integrations.
  • <span>Terminal output control</span>: <span>Roo Cline</span> allows developers to control the number of terminal output lines passed to the model during command execution, which is very useful for debugging.

Performance and Cost Optimization

  • <span>Lower token consumption</span>: <span>Roo Cline</span> may reduce token consumption during interactions through caching or batch operations, saving development costs.
  • <span>MCP protocol optimization</span>: <span>Roo Cline</span> supports custom MCP (<span>Model Context Protocol</span>) settings, including automatic retry mechanisms and automatic approval for high-risk operations, improving development efficiency.

User Experience Improvements

  • <span>Drag and drop image and screenshot function</span>: <span>Roo Cline</span> supports dragging images into the chat area and adjusting screenshot quality, allowing developers to quickly generate or modify interfaces.
  • <span>History management</span>: <span>Roo Cline</span> provides more convenient history management features, supporting one-click copying of prompts and message deletion.
  • <span>Browser interaction</span>: <span>Roo Cline</span> can directly launch a browser for interactive testing, supporting clicks, text input, scrolling, etc., and capturing screenshots and console logs.

Overall, <span>Roo Cline</span> is a comprehensive optimization based on <span>Cline</span>, providing richer features, lower costs, and better user experience.

For developers, <span>Roo Cline</span> is a more flexible and efficient choice, especially suitable for scenarios that require frequent model switching or complex debugging.

If you are already familiar with <span>Cline</span>, <span>Roo Cline</span> will bring more surprises.

Transforming Programming Efficiency with DeepSeek R1 in VSCode
Immediate Support for DeepSeek R1

On the second day after the release of <span>deepseek r1</span>, <span>Roo Cline</span> has already supported the <span>deepseek r1</span> model.

Transforming Programming Efficiency with DeepSeek R1 in VSCode
Extension Settings

After installing the <span>Roo Cline</span> extension in VSCode, configure it as shown in the image above. For convenience, the parameters are as follows:

  1. 1. <span>Select DeepSeek</span>
  2. 2. <span>No need for https://api.deepseek.com</span>
  3. 3. Your <span>api key</span>, find it in the DeepSeek user backend.
  4. 4. <span>deepseek-reasoner</span>

After this setup, start enjoying the programming efficiency brought by <span>deepseek r1</span>.

DeepSeek backend address: <span>https://platform.deepseek.com</span>.

Transforming Programming Efficiency with DeepSeek R1 in VSCode
My Usage

I recharged 20 Yuan on December 30, and almost a month later, I still have 19.64 Yuan left.

On one hand, I really haven’t used it frequently, and on the other hand, DeepSeek’s pricing is indeed friendly.

I used to scoff at AI programming, but after experiencing it, it truly opened my eyes; the power of technology is like magic.

Let’s embrace AI together and become super programming warriors with ten times the power.

Previous articles:

  • Rural Programmer: Save 10,000, Return to the Village, Save 10,000, Start a Business
  • Rural Programmer’s 2024 Year-End Summary, Struggles, Confusion, and Hope
  • Frontend Development is Really Hard
  • 2025 Frontend First Battle, Those Who Write Vue are Crazy, Even the Ancestor of Vue is Here
  • Rural Programmer: You Have Too Many Things, Not Focused Enough
  • Rural Programmer: House, Car, Child, Education, Career, and Retirement
  • Rural Programmer: Legendary Past, Dropped Out in Sophomore Year with 8 Fails, Repeated High School Twice
  • VSCode v1.96 Released, AI Significantly Enhanced, Experience Greatly Improved
  • Node.js Competitor Bun is at it Again, Bun.s3 Preview
  • A 10x Cheaper Alternative to Cursor has Arrived
  • Independent Development: Where Do Ideas and Creativity Come From?
  • Independent Development: The Value of Freedom
  • Independent Development: Desires and Ambitions
  • Independent Development: What to Do When Your Mind Breaks?
  • Independent Development: Discussing Costs
  • pnpm v10 Official Release, Major Update, Lasting 3 Months, 12 Versions
  • If You Want to Earn Money Beyond Your Salary, What Should You Do?
  • Free HTTPS Management, Monitoring, and Auto-Renewal Tool – httpsok
  • Suiyi Weekly – Issue 007

Leave a Comment