How to Run Cline with Berget AI in Visual Studio Code

Cline is a powerful AI coding assistant that can help you write, refactor, and improve your code directly in Visual Studio Code. In this guide, we'll show you how to configure Cline to use Berget AI's models, giving you access to European AI models with high performance for code generation.

What is Cline?

Cline (formerly known as Claude Dev) is a Visual Studio Code extension that integrates AI-assisted coding directly into your development environment. Unlike other AI coding tools, Cline can:

Read and write files in your project
Run terminal commands to test and build code
Understand project context by analyzing the entire codebase
Perform complex refactoring across multiple files
Debug and fix problems automatically

Why Use Berget AI with Cline?

The Problem with Foreign AI Coding Tools

Many developers today use tools like Claude Code, GitHub Copilot, or Visual Studio IntelliCode without considering where their code and business secrets end up. When you use these services, your code is sent to servers in the US or other countries, creating several problems:

Business secrets at risk: Your proprietary code, API keys, and business logic are sent to foreign servers
GDPR violations: Personal data in code may be processed outside the EU without adequate protection
Corporate policy violations: Many organizations have policies prohibiting sensitive code from leaving the company
Legal uncertainty: The Cloud Act and FISA give US authorities the right to request data from US companies
No control: You don't know who has access to your code or how it's used to train future models

Berget AI - The Swedish Alternative

Berget AI solves these problems by offering several advantages for code generation:

European AI sovereignty: Your code and data stays within the EU under European legislation
Specialized code models: GLM-4.6, Qwen 3.2 32B, and Magistral are optimized for code generation
Transparent pricing: No hidden costs or surprises
High performance: Fast responses and low latency from Swedish data centers
No training data: Your code is never used to train models or improve the service
Full control: You retain full control over your code and can be sure it never leaves Sweden

Step 1: Install Cline in Visual Studio Code

Start by installing the Cline extension:

Open Visual Studio Code
Go to the Extensions panel (Ctrl+Shift+X)
Search for "Cline"
Click "Install" for the Cline extension

Installing the Cline extension in VS Code

Step 2: Create an API Key from Berget AI

You need an API key to use Berget AI's models. There are two ways to create one:

Option A: Via CLI Tool

If you have Berget AI's CLI installed:

# Log in to your account
berget auth login

# Create a new API key
berget api-keys create --name "Cline Development"

Create API key with Berget CLI

Option B: Via Console

Go to console.berget.ai
Log in to your account
Navigate to "API Keys" in the sidebar
Click "Create New Key"
Give the key a descriptive name (e.g., "Cline VS Code")
Copy the generated key

Create API key in Berget AI Console

Important: Save your API key securely - it's only shown once!

Step 3: Configure Cline with Berget AI

Now let's configure Cline to use Berget AI's API:

Open the Cline panel in VS Code (click the Cline icon in the sidebar)
Click "Configure API" or the settings icon
Select "Custom API" as provider
Fill in the following information:

API Configuration:

Base URL: https://api.berget.ai/v1
API Key: Your API key from step 2
Model: glm-4.6 (recommended for code generation)

Configuration of Cline with Berget AI's API

Step 4: Choose the Right Model for Code Generation

Berget AI offers several models that work excellently with Cline:

GLM-4.6 (Recommended for code generation)

Best for: Code generation, refactoring, debugging, complex programming tasks
Strengths: Excellent understanding of programming context, strong performance on HumanEval benchmark
Language support: Supports all popular programming languages including Python, JavaScript, TypeScript, Go, Rust
Performance: High quality code generation, good at following instructions
Model ID: glm-4.6

Qwen 3.2 32B

Best for: Complex code projects, architectural decisions, code review
Strengths: Very strong code understanding, excellent at reasoning and problem-solving
Language support: Broad language support with particularly strong performance on Python and JavaScript
Performance: Top performance on code-related benchmarks, good balance of speed and quality
Model ID: qwen-3.2-32b

Magistral Small

Best for: Quick code suggestions, smaller refactoring, prototyping
Strengths: European model with good code understanding, fast responses
Advantages: Lower latency, cost-effective, good for iterative development
Performance: Solid performance for smaller coding tasks, optimized for speed
Model ID: magistral-small

AI model selection in Cline

Step 5: Test Your Configuration

Let's test that everything works:

Open a code project in VS Code
Open the Cline panel

Write a simple request, for example:

Create a simple HTTP server in Node.js that responds with "Hello World" on port 3000

Cline should now generate code and possibly create files for you

First test of Cline with Berget AI

Advanced Configurations

Customize Model Parameters

You can fine-tune the model's behavior by adding parameters to the configuration:

{
  "baseUrl": "https://api.berget.ai/v1",
  "apiKey": "your-api-key",
  "model": "glm-4.6",
  "temperature": 0.1,
  "maxTokens": 4096
}

Project-Specific Settings

For different projects, you might want to use different models based on your needs:

Large, complex projects: Qwen 3.2 32B for deep code understanding and architectural decisions
General code generation: GLM-4.6 for balanced performance and quality
Rapid prototyping: Magistral Small for fast iterations and cost-effectiveness
Production code: GLM-4.6 or Qwen 3.2 32B with low temperature (0.1) for deterministic results
Code review: Qwen 3.2 32B for in-depth analysis and improvement suggestions
Bug fixing: GLM-4.6 for efficient debugging and problem-solving

Performance Comparison

Based on benchmarks and real-world usage:

Model	Code Quality	Speed	Cost	Best For
Qwen 3.2 32B	⭐⭐⭐⭐⭐	⭐⭐⭐	⭐⭐⭐	Complex projects
GLM-4.6	⭐⭐⭐⭐	⭐⭐⭐⭐	⭐⭐⭐⭐	General coding
Magistral Small	⭐⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐⭐⭐	Quick tasks

Practical Tips for Using Cline Effectively

1. Be Specific in Your Requests

❌ "Fix this function"
✅ "Refactor the getUserData function to use async/await instead of callbacks"

2. Give Context About Your Project

✅ "In this React project with TypeScript, create a new component for user profile page"

3. Use Cline for Entire Workflows

✅ "Create a new API endpoint, add tests, and update the documentation"

4. Let Cline Run Tests

Cline can run your tests and fix problems automatically:

✅ "Run the tests and fix any errors that occur"

Troubleshooting

Problem: "API Key Invalid"

Check that you copied the entire API key correctly
Verify that the key hasn't expired
Test the key with a simple curl command:

curl -H "Authorization: Bearer your-api-key" \
     -H "Content-Type: application/json" \
     https://api.berget.ai/v1/models

Problem: "Model Not Found"

Check that the model name is spelled correctly
Use glm-4.6, qwen-3.2-32b or magistral-small
Verify that your API key has access to the model

Problem: Slow Responses

Try switching to magistral-small for faster responses
Reduce maxTokens in the configuration
Check your internet connection

Common troubleshooting steps for Cline

Security and Best Practices

Protect Your API Key

Never store API keys in code that's committed to Git
Use environment variables or VS Code's secure storage
Rotate keys regularly

Review Generated Code

Cline is powerful but not infallible
Always review code before running it in production
Use your usual code review processes

Limit Access

Create separate API keys for different projects
Use keys with limited permissions when possible

Conclusion

With Cline and Berget AI, you get a powerful AI coding assistant that:

Respects your privacy through European AI sovereignty
Delivers high quality with specialized code models
Integrates seamlessly into your existing development environment
Scales with your needs from prototypes to production

GLM-4.6 is our recommendation for general code generation, Qwen 3.2 32B for complex projects requiring deep understanding, while Magistral Small is perfect when you need fast responses for smaller tasks.

Get started today and experience how AI-assisted coding can accelerate your development!

Need help with configuration or have questions about our models? Contact us or visit our documentation.