Content
# 🚀 Free AI Coding with Qwen 3.8-Max + MCP
> **Code for free** using one of the most powerful open-source AI models — no API keys, no subscriptions.
This guide shows you how to set up **[Qwen 3.8-Max](https://qwen.ai/blog?id=qwen3.8)** (via Qwen Studio) with MCP (Model Context Protocol) to read, write, and edit files on your machine — turning it into a fully capable AI coding assistant.
> ⚠️ **Important:** Qwen 3.8-Max runs **in the cloud** through Qwen Studio. Your prompts and any file contents read via MCP **are sent to Qwen's servers** for processing. The MCP servers themselves (Filesystem, Fetch, etc.) run locally on your machine. Review [Qwen's privacy policy](https://qwen.ai) before using this with sensitive or proprietary code.
---
## 🎬 Demo
<p align="center">
<img src="assets/demo.gif" alt="Setting up Qwen 3.8-Max with MCP for coding" width="720" />
</p>
---
## ✨ Why This Setup?
| Feature | Details |
|---|---|
| 💰 **Cost** | Free to use (no API keys or subscriptions needed) |
| ☁️ **How it works** | Model runs in the cloud via Qwen Studio; MCP servers run locally on your machine |
| ♾️ **Usage** | Free tier — subject to [Qwen's usage policies](https://qwen.ai) and may vary |
| 🧠 **Model** | [Qwen 3.8-Max](https://qwen.ai/blog?id=qwen3.8) — 2.4T parameter MoE model (~95B active), 1M context window, native multimodal |
| 🔌 **Integration** | MCP Filesystem — the model can read/write your local files via MCP |
---
## 🛠️ Step-by-Step Setup
### Step 1: Download & Install Qwen Studio
1. Go to **[qwen.ai/download](https://qwen.ai/download)** and download the desktop application for your operating system (available for **macOS** and **Windows** only).
2. Install and open Qwen Studio.
### Step 2: Open MCP Settings
1. In the Qwen Studio interface, look for the **Add (+)** icon or the **Settings (⚙️)** icon.
2. Click on it to open the settings panel.
3. Find and click on the **MCP** option.
### Step 3: Enable Filesystem MCP
1. Navigate to **"My MCPs"** section.
2. Locate the MCP called **"Filesystem"**.
3. **Enable it** by toggling it on.
### Don't see it pre-listed?
Qwen Studio fetches MCPs at runtime, so you can add any public MCP manually via **Add MCP**.
#### 📁 Filesystem (Node/npm — runs via `npx`)
> ⚠️ Requires **[Node.js](https://nodejs.org/)** installed — `npx` comes bundled with it.
**Option 1: JSON Configuration**
To give access to multiple specific folders, add each path as a new string in the `args` array (separate them with commas).
```json
{
"mcpServers": {
"filesystem": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem@latest",
"/path/to/folder1",
"/path/to/folder2",
"/path/to/folder3"
]
}
}
}
```
**Option 2: Via the Form**
1. Click **Add MCP** → **Quick Add**.
2. **Name:** Enter a simple, lowercase identifier with no spaces (e.g., `filesystem` or `my-files`).
3. **Description:** Optional (e.g., "Access to project files").
4. **Type:** Choose **STDIO (standard input/output)**.
5. **Command:** Choose `npx`.
6. **Parameters:** Paste the config below. To add multiple folders, separate the paths with **spaces**:
`-y @modelcontextprotocol/server-filesystem@latest /path/to/folder1 /path/to/folder2`
---
### Step 4: Start Coding!
That's it! You now have a fully functional AI coding assistant that can:
- 📖 **Read** your project files and understand your codebase
- ✏️ **Write** new files and create entire projects
- 🔧 **Edit** existing code with context-aware modifications
- 📁 **Navigate** your file system and understand project structure
---
## 💡 Tips for Best Results
- **Be specific** — Tell the model exactly which files or directories to work with
- **Provide context** — Mention the programming language, framework, or library you're using
- **Iterate** — Start with small tasks and build up to larger ones
- **Review changes** — Always review the code changes before accepting them
---
## 🤔 FAQ
<details>
<summary><strong>Is this really free?</strong></summary>
Yes! Qwen Studio provides free access to Qwen 3.8-Max — no API keys or subscriptions required. The model runs in the cloud on Qwen's infrastructure. Usage may be subject to Qwen's terms and fair-use policies.
</details>
<details>
<summary><strong>What is Qwen 3.8-Max?</strong></summary>
Qwen 3.8-Max is Alibaba's most capable open-source model, released on August 3, 2026. It's a 2.4-trillion-parameter sparse Mixture-of-Experts (MoE) model with ~95 billion active parameters per token. It supports a 1-million-token context window and native multimodal input (text, image, video). Read the [official blog post](https://qwen.ai/blog?id=qwen3.8) for full details.
</details>
<details>
<summary><strong>How does MCP work?</strong></summary>
MCP (Model Context Protocol) is an open standard that allows AI models to interact with external tools and data sources. The Filesystem MCP gives the model permission to read and write files on your local machine, enabling it to act as a coding assistant.
</details>
<details>
<summary><strong>Is my code sent to the cloud?</strong></summary>
Yes. Qwen 3.8-Max runs on Qwen's cloud servers — your prompts and any file contents read by the MCP Filesystem **are sent to Qwen's servers** for inference. The MCP servers themselves run locally on your machine, but the model processing happens in the cloud. Avoid using this with sensitive, proprietary, or confidential code unless you've reviewed and accepted [Qwen's privacy policy](https://qwen.ai).
</details>
<details>
<summary><strong>What kind of coding tasks can it handle?</strong></summary>
Pretty much anything — writing new code, debugging, refactoring, explaining code, creating project structures, writing tests, and more. Qwen 3.8-Max excels at coding, professional "cowork" tasks, and long-horizon agentic workflows.
</details>
---
## 📚 Resources
- 📥 [Download Qwen Studio](https://qwen.ai/download)
- 📝 [Qwen 3.8-Max Blog Post](https://qwen.ai/blog?id=qwen3.8)
- 🌐 [Qwen Official Website](https://qwen.ai)
---
## 🌟 Star This Repo
If you found this useful, please ⭐ **star this repo** to help others discover it!
---
## 📄 License
This project is licensed under the [MIT License](LICENSE).
