From Prompt to Production: Baseten Inference in Your IDE with Cline

Get Started with Kimi K2, Qwen3 Coder, and DeepSeek R1 via Baseten with Cline

Cline + Baseten

Cline is an open-source AI coding assistant that transforms your development workflow by bringing powerful language models directly into your favorite IDE. With our new Cline integration, you can now leverage cutting-edge models like DeepSeek R1, Kimi K2, and Qwen3 Coder to accelerate your coding projects with enterprise-grade performance and reliability.

This integration brings together Cline's intuitive coding interface with Baseten's robust model serving infrastructure, giving you access to state-of-the-art open-source models optimized for different coding workflows. Whether you're building complex applications, debugging code, or exploring new technologies, this combination provides the tools you need for modern software development.

Prerequisites

Before starting, you'll need:

  1. Cline installed in your IDE. Add Cline to your development environment by visiting cline.bot and following the installation instructions for your preferred IDE.

  2. A Baseten account for Model APIs access. Sign up at baseten.co to get started. 

Configuring Baseten in Cline

Step 1: Open your IDE and activate the primary sidebar (you can use the Command+B shortcut on macOS). 

Open primary sidebarOpen primary sidebar

Step 2: Click the up arrow to expand the sidebar options and select the Cline icon to open the assistant interface.

Expand options and click ClineExpand options and click Cline

You should see a window like this open.

Cline interfaceCline interface


Step 3. Click the gear icon in the Cline interface to access configuration options. From the API Provider dropdown, select "Baseten" as your model provider. Navigate to app.baseten.co/settings/api_keys to generate your API key and copy it into the Baseten API Key field. You can select from the model dropdown your preferred model to code with. We recommend Qwen3 Coder for this tutorial. Then click “Done” to save the settings.

Select Baseten as provider, supply API key, and choose modelSelect Baseten as provider, supply API key, and choose model

Building Flappy Bird: A Complete Example

You can now start prompting the AI in either Plan or Act mode. For example, we can use the prompt “generate flappy bird in one file and launch the game!” to one-shot a version of Flappy Bird you can play in your browser.

In the Cline interface, you can choose between "Plan" and "Act" modes:

  • Plan mode will gather information to architect a plan

  • Act mode immediately begins implementation based on your prompt

For our Flappy Bird example, try this prompt under act mode.

`Generate flappy bird in one file and launch the game!`

Watch as the AI assistant:

  1. Analyzes the requirements for a Flappy Bird implementation

  2. Generates complete HTML, CSS, and JavaScript code

  3. Creates game mechanics including physics, collision detection, and scoring

  4. Produces a fully playable browser-based game

Sample prompt to generate Flappy BirdSample prompt to generate Flappy Bird
Flappy Bird in your browser created by Baseten + ClineFlappy Bird in your browser created by Baseten + Cline

Choose the Right Model for Your Project

Different coding scenarios benefit from different models. For rapid prototyping and quick implementations, Qwen3 Coder's speed delivers fast iterations and immediate feedback. For complex system integrations, Kimi K2's function calling capabilities excel when working with APIs, webhooks, or multi-service architectures. For comprehensive analysis and reasoning-heavy tasks, DeepSeek R1 provides thoughtful code reviews, architectural suggestions, and detailed problem-solving across multiple programming languages.

Feel free to experiment with our full list of supported models—all accessible with a single Baseten API key. 

Whether you're building a weekend prototype or enterprise software, Baseten's robust, lightning-fast inference infrastructure can power your development workflow through Cline. We can't wait to see what you create. 

Subscribe to our newsletter

Stay up to date on model performance, GPUs, and more.