cerebras

Cerebras Cloud (WSE-3 Inference)

Cerebras Cloud provides free access to flagship open-weight models using their Wafer-Scale Engine (WSE-3), offering high parallelization and token throughput.

🚀 Key Specs (2026)

• Model Support: GPT-OSS 120B, Llama 3 series, specialized engineering models.
• Hardware Tier: Wafer-Scale Engine (WSE-3) architecture.
• Rate Limit (Free Tier):
  • Daily Request Cap: 14,400 RPD.
  • Requests per Minute: 30 RPM.
  • Tokens per Minute: 60,000 TPM (Higher throughput for batch tasks).
• Primary Strength: High concurrency and batch processing.

🗝 How to get an API Key

1. Visit Cerebras Cloud Console.
2. Create an account.
3. Generate a key from the “API Keys” section.

🛠 Usage Example (Python)

from openai import OpenAI
 
client = OpenAI(
    base_url="https://api.cerebras.ai/v1",
    api_key="YOUR_CEREBRAS_API_KEY",
)
 
response = client.chat.completions.create(
    model="llama-3.1-70b",
    messages=[{"role": "user", "content": "What is parallel computing?"}]
)
print(response.choices[0].message.content)

🔗 Official Links

• Cerebras Cloud
• Documentation