top of page
Search

Qwen 3: A Powerful Open-Source Coding Assistant in 2025

  • Philip Moses
  • 2 days ago
  • 4 min read

Updated: 21 hours ago

Coding assistants are getting more popular, thanks to tools like Claude Code and OpenAI Codex CLI. Now, there’s a new player in the game—Qwen 3. It’s a free, open-source AI model designed to help with coding, whether you're stuck on a tricky problem or just want a smarter way to write code.

In this blog, we’ll explore what makes Qwen 3 special and how you can start using it.

 

What is Qwen3-Coder?

Qwen3-Coder is the latest AI model from the Qwen team. The most powerful version, Qwen3-Coder-480B-A35B-Instruct, has a massive 480 billion parameters. But don’t worry—it doesn’t use all of them at once!


Key Features:

480 Billion Parameters – Only 35 billion are active at a time, making it efficient.

Mixture-of-Experts (MoE) – Instead of using all its brainpower for every task, it activates only the "experts" needed.

Huge Memory (256K Tokens) – It can handle long code files and even entire projects.

Scales Up to 1 Million Tokens – With extrapolation, it can process even larger inputs.

This makes Qwen3-Coder great for understanding big codebases and complex tasks.


How Does Qwen3-Coder Work?

The model is built for coding tasks, from writing new code to fixing bugs. Here’s how it’s designed:

  • MoE Architecture – It has 160 experts, but only 8 are active at a time, balancing power and speed.

  • Trained on 7.5 Trillion Tokens – 70% of that was code, so it really understands programming.

  • Fine-Tuned with Human Feedback – This helps it handle real-world coding challenges better.


The graph below shows how Qwen3-Coder keeps improving across different coding tasks:

Source: GitHub
Source: GitHub
Performance of Qwen3-Coder

Qwen3-Coder achieved a state-of-the-art agentic performance in comparison to other open-source models on the SWE-Bench benchmark. As shown in the graph, it achieves 69.6% verified accuracy in a 500-turn interactive setting and 67.0% in single-shot mode. It outperformed other models like Mistral-small-2507 with 53.6% and GPT-4.1 with 54.6% accuracy. It ranks just behind Claude-Sonnet-4 (70.4%) and ahead of Kimi-K2 (65.4%), and Gemini-2.5 (49.0%). This establishes Qwen3-Coder as the top-performing open agentic model for real-world software engineering tasks.


Source: GitHub
Source: GitHub

Getting Started with Qwen Code

To access Qwen Code directly, head over to https://chat.qwen.ai/, and there you can select Qwen3-Coder as the model and start using it.

ree

Qwen API

You can directly access the API of Qwen3-Coder through Alibaba Cloud Model Studio. Here is a demonstration of how to use this model with the Qwen API. As of now, no free quota is available.


import os

from openai import OpenAI


# Create client - using intl URL for users outside of China

# If you are in mainland China, use the following URL:

# "https://dashscope.aliyuncs.com/compatible-mode/v1"


client = OpenAI(

api_key=os.getenv("DASHSCOPE_API_KEY"),

base_url="https://dashscope-intl.aliyuncs.com/compatible-mode/v1",

)


prompt = "Help me create a web page for an online bookstore."


# Send request to qwen3-coder-plus model

completion = client.chat.completions.create(

model="qwen3-coder-plus",

messages=[

{"role": "system", "content": "You are a helpful assistant."},

{"role": "user", "content": prompt}

],

)


# Print the response

print(completion.choices[0].message.content.strip())

The Qwen team has also released a command-line tool called Qwen Code to make it easy to use Qwen3-Coder. Here is a step-by-step guide to get you started:


How to Use Qwen Code?

Step 1: Install Node.js

First, you will need to install Node.js version 20 or higher on your system. You can install it with the following commands. Open your terminal and paste the following commands one by one.


# Download and install nvm:


curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh | bash


# in lieu of restarting the shell


\. "$HOME/.nvm/nvm.sh"


# Download and install Node.js:


nvm install 22


# Verify the Node.js version:


node -v # Should print "v22.17.1".


nvm current # Should print "v22.17.1".


# Verify npm version:


npm -v # Should print "10.9.2".

Step 2: Install Qwen Code

Next, install the Qwen Code tool using the npm package manager: 

npm i -g @qwen-code/qwen-code

You should see something like this:

ree

Step 3: Configure Your API Key

You can directly access the API of Qwen3-Coder through Alibaba Cloud Model Studio. As of now, no free quota is available.

You will need to set up your API key to use the model. You can do this by setting environment variables. 

export OPENAI_API_KEY="your_qwen_api_key_here"

export OPENAI_BASE_URL="https://dashscope-intl.aliyuncs.com/compatible-mode/v1"

export OPENAI_MODEL="qwen3-coder-plus"

Step 4: Start Coding

Now you are ready to use `Qwen Code`. You can navigate to your project directory and start interacting with the agent. For example, to understand the architecture of a project, you can use the command or just write the following command qwen code will pop up:

qwen

ree

You can also use it for more complex tasks like refactoring code or even automating workflows.


How to Use Qwen3-Coder in Claude Code?

In addition to Qwen Code, you can now use Qwen3‑Coder with Claude Code. Simply request an API key on Alibaba Cloud Model Studio platform and install Claude Code to start coding.

npm install -g @anthropic-ai/claude-code Set up environment variables for using Qwen3‑Coder export ANTHROPIC_BASE_URL=https://dashscope-intl.aliyuncs.com/api/v2/apps/claude-code-proxy export ANTHROPIC_AUTH_TOKEN=your-dashscope-apikey

Then you should be able to use Claude Code with Qwen3-Coder!

Note: You can use either Qwen CLI or Web Interface to perform coding tasks.

Conclusion

Qwen3-Coder represents an enormous breakthrough in open-source AI models within the domain of code generation. Its powerful architecture, massive context window, and agentic capabilities make it a valuable tool for developers and researchers. Because the model continues to be developed, we will expect to see even more impressive features and performance in the future. This open-source AI model is set to have a significant impact on how we approach software development problems, making it more efficient and automated.

 
 
 

Recent Posts

See All

Comments


Curious about AI Agent?
bottom of page