Cerebras Coder

An AI code assistant powered by Cerebras that generates fully functional web applications in seconds.

Inspired by Hassan's Llama Coder.

Features

Multiple Model Support: Choose from three powerful AI models:
- Qwen 3 235B Instruct (235B parameters, ~1400 tokens/s)
- Z.ai GLM 4.6 (357B parameters, ~1000 tokens/s)
- OpenAI GPT OSS (120B parameters, ~3000 tokens/s)
Flexible API Key Management:
- Use the default hosted API key (no setup required)
- Or use your own personal Cerebras API key for private usage
Instant Code Generation: Turn your ideas into fully functional apps in less than a second
Live Preview: See your code come to life in real-time
Version History: Track and revert to previous versions of your code
Share Your Projects: Generate shareable links to your creations

Simply open the application and start coding! The default API key is already configured.

Note: Personal API keys are stored only in browser memory and are forgotten when you close or refresh the page.

To deploy with your own default API key:

Save your Cerebras API key in your project environment variable called CEREBRAS_API_KEY
The application will use this key by default

Select a Model: Click the model dropdown in the header to choose your preferred AI model
Choose API Key Mode: Select between "Default API Key" or "Personal API Key"
Enter Your Prompt: Type what you want to build (e.g., "Build me a calculator app")
Generate Code: Click the prompt button or press Enter
View & Edit: See your code in the editor and preview it live
Iterate: Make changes by entering new prompts
Share: Click the Share button to get a link to your project

Model	Parameters	Speed	Best For
Qwen 3 235B Instruct	235B	~1400 tokens/s	Balanced performance
Z.ai GLM 4.6	357B	~1000 tokens/s	Complex reasoning
OpenAI GPT OSS	120B	~3000 tokens/s	Fast generation

This project is open source. Feel free to fork and customize it for your needs!