@ai-sdk/cerebras

AI SDK - Cerebras Provider

The Cerebras provider for the AI SDK contains language model support for Cerebras, offering high-speed AI model inference powered by Cerebras Wafer-Scale Engines and CS-3 systems.

Setup

The Cerebras provider is available in the @ai-sdk/cerebras module. You can install it with

npm i @ai-sdk/cerebras

Provider Instance

You can import the default provider instance cerebras from @ai-sdk/cerebras:

import { cerebras } from '@ai-sdk/cerebras';

Available Models

Cerebras offers a variety of high-performance language models:

Llama 3.3 70B

• Model ID: llama-3.3-70b
• 70 billion parameters
• Knowledge cutoff: December 2023
• Context Length: 8192
• Training Tokens: 15 trillion+

Llama 3.1 8B

• Model ID: llama3.1-8b
• 8 billion parameters
• Knowledge cutoff: March 2023
• Context Length: 8192
• Training Tokens: 15 trillion+

GPT-OSS 120B

• Model ID: gpt-oss-120b
• 120 billion parameters
• High-performance open-source model
• Optimized for inference speed

Qwen 3 235B A22B Instruct 2507

• Model ID: qwen-3-235b-a22b-instruct-2507
• 235 billion parameters
• Instruction-tuned model
• Released July 2025

Qwen 3 235B A22B Thinking 2507

• Model ID: qwen-3-235b-a22b-thinking-2507
• 235 billion parameters
• Enhanced reasoning capabilities
• Released July 2025

Qwen 3 32B

• Model ID: qwen-3-32b
• 32 billion parameters
• Balanced performance and efficiency
• Multilingual capabilities

Qwen 3 Coder 480B

• Model ID: qwen-3-coder-480b
• 480 billion parameters
• Specialized for code generation and understanding
• Advanced programming capabilities

Example

import { cerebras } from '@ai-sdk/cerebras';
import { generateText } from 'ai';

const { text } = await generateText({
  model: cerebras('llama-3.3-70b'),
  prompt: 'Write a JavaScript function that sorts a list:',
});

Documentation

For more information about Cerebras' high-speed inference capabilities and API documentation, please visit:

• Cerebras Inference Documentation
• Cerebras Website

Note: Due to high demand in the early launch phase, context windows are temporarily limited to 8192 tokens in the Free Tier.