English

Overview

Cerebras Systems builds the world's largest computer chip - the Wafer Scale Engine (WSE) - designed specifically for AI workloads. This cookbook provides comprehensive examples, tutorials, and best practices for developing and deploying AI models using Cerebras infrastructure, including both training on WSE clusters and fast inference via Cerebras Cloud.

What You'll Find in This Repository

Section	Description
Get Started	SDK setup guides, authentication, and model exploration examples
Agents	AI agent implementations using Cerebras with CrewAI and Agno frameworks
Chat with Data	RAG implementations, multilingual PDF chat, and fusion techniques
Integrations	Framework integrations including LiteLLM for seamless API switching
Starter Apps	Production-ready applications: AI book writer and mindmap generator

Cerebras Starter Guide

Get Your API Key

Get your API key by visiting Cerebras Cloud.

⚠️ Security Note: Keep your API key secure! Never expose it in client-side code or public repositories.

API Endpoint

https://api.cerebras.ai/v1/chat/completions

Quick Start

Prerequisites

• A Cerebras account
• A Cerebras Inference API key
• Python 3.7+ or Node.js 14+

Set up your API key

export CEREBRAS_API_KEY="your-api-key-here"

Install the Cerebras SDK

pip install --upgrade cerebras_cloud_sdk

Sample Code

Python

import os
from cerebras.cloud.sdk import Cerebras

client = Cerebras(
    api_key=os.environ.get("CEREBRAS_API_KEY"),
)

chat_completion = client.chat.completions.create(
    messages=[
        {
            "role": "user",
            "content": "Explain the advantages of wafer-scale computing for AI",
        }
    ],
    model="llama-4-scout-17b-16e-instruct",
)

print(chat_completion.choices[0].message.content)

cURL

curl -X POST "https://api.cerebras.ai/v1/chat/completions" \
  -H "Authorization: Bearer $CEREBRAS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "llama-4-scout-17b-16e-instruct",
    "messages": [
      {"role": "user", "content": "What makes Cerebras WSE unique for AI training?"}
    ],
    "max_tokens": 1024,
    "temperature": 0.7
  }'

JavaScript/Node.js

import { Cerebras } from 'cerebras_cloud_sdk';

const client = new Cerebras({
    apiKey: process.env.CEREBRAS_API_KEY,
});

async function main() {
    const chatCompletion = await client.chat.completions.create({
        messages: [
            {
                role: 'user',
                content: 'How does wafer-scale architecture improve AI performance?',
            },
        ],
        model: 'llama-4-scout-17b-16e-instruct',
    });

    console.log(chatCompletion.choices[0].message.content);
}

main();

Features

• Fast Inference: Ultra-low latency inference with Cerebras Cloud
• Model Zoo Integration: Access to optimized pre-trained models
• OpenAI-compatible API: Easy migration from OpenAI to Cerebras
• Streaming Support: Real-time response streaming capabilities
• Enterprise Ready: Production-grade infrastructure and support

Available Models

Production Models

Model	Model ID	Description	Use Cases
Llama 4 Scout 17B	llama-4-scout-17b-16e-instruct	High-performance instruction-tuned model	Chat, reasoning, code generation
Llama 3.1 8B	llama3.1-8b	Efficient general-purpose model	Text generation, summarization
Llama 3.3 70B	llama-3.3-70b	Large-scale model for complex tasks	Advanced reasoning, research
OpenAI GPT OSS 120B	gpt-oss-120b	Open-source GPT-style model	General text generation, chat
Qwen 3 32B	qwen-3-32b	Multilingual model with strong reasoning	Multilingual tasks, reasoning

Preview Models

Model	Model ID	Description	Use Cases
Llama 4 Maverick 17B	llama-4-maverick-17b-128e-instruct	Extended context Llama 4 variant	Long-form content, extended reasoning
Qwen 3 235B Instruct	qwen-3-235b-a22b-instruct-2507	Large instruction-tuned model	Complex reasoning, advanced tasks
Qwen 3 235B Thinking	qwen-3-235b-a22b-thinking-2507	Reasoning-focused variant	Chain-of-thought, problem solving
Qwen 3 480B Coder	qwen-3-coder-480b	Specialized code generation model	Code generation, programming tasks

Resources

• Documentation: Cerebras Docs
• Inference API: Inference Documentation
• Training API: Training Documentation
• Model Zoo: Cerebras Model Zoo
• SDK: Python SDK | Node.js SDK
• Playground: Try Cerebras Cloud
• Live Demo: Cerebras Inference Demo

Getting Support

• Issues: Report bugs or request features via GitHub Issues
• Documentation: Visit Cerebras Documentation
• Contact: Reach out to Cerebras support team

Contributing

We welcome contributions to the Cerebras-Cookbook repository! To contribute:

1. Fork and clone the repository
2. Create a new branch for your changes
3. Make your changes following our documentation standards
4. Test your examples thoroughly
5. Submit a pull request

For major changes, please open an issue first to discuss your ideas.

Contribution Guidelines

• Include clear documentation and comments
• Provide example usage and expected outputs
• Follow the existing code style and structure
• Add appropriate tests where applicable
• Update the README if adding new sections

Legal

• Privacy Policy
• Terms of Service

---

© 2025 Cerebras Systems | All Rights Reserved