Cerebrium

Verified 💳 Trial Credits

Serverless GPU infrastructure for deploying your own AI models with per-second billing and autoscaling. The Hobby plan has no monthly fee, but compute is always billed; Cerebrium's pricing page no longer shows a sign-up credit. Useful for hosting a model you choose, not a free shared API.

$30 Credits Serverless GPU Custom Deploy Auto-Scaling
Get API Key Suggest Edit
0

Overview

Provider Type

Trial Credits

API Endpoint

https://api.cortex.cerebrium.ai/v4

Free Tier Highlights

Pay-per-second compute

Why Choose Cerebrium?

Cerebrium is serverless GPU hosting: you deploy the model of your choice and pay per second of compute. The Hobby plan has no monthly fee.

Best for: hosting your own or a fine-tuned model without managing servers.

Watch out: it is not a free shared API. Compute is always billed, and the pricing page no longer shows a sign-up credit.

Quick Start Guide

1

Visit https://www.cerebrium.ai/

2

Sign up for free account

3

Receive $30 free credits

4

Install cerebrium CLI: pip install cerebrium

5

Deploy model with cerebrium deploy

6

Access via generated API endpoint

Available Models

2 free models
Model Name ID Context
Any HuggingFace Model
custom-deployment
Model-dependent
Llama 3.1 (Deployable)
meta-llama/Llama-3.1-8B-Instruct
128 000

Integration Examples

Ready-to-use code snippets for your applications.

main.py
# Install: pip install cerebrium
from cerebrium import get_secret

# Cerebrium uses Python-native deployments
# main.py (deployed to Cerebrium)
from vllm import LLM, SamplingParams

llm = LLM(model="meta-llama/Llama-3.1-8B-Instruct")

def predict(prompt: str):
    params = SamplingParams(temperature=0.7, max_tokens=512)
    output = llm.generate([prompt], params)
    return {"response": output[0].outputs[0].text}

# Deploy: cerebrium deploy my-llm

Free Tier Pricing & Limits

Rate Limit

Requests per minute

Pay-per-second compute

Daily Quota

Requests per day

Credit-based

Token Limit

Tokens per minute

$30 free credits

Monthly Quota

Per month limit

One-time $30 credit

Free Credits

One-time

No sign-up credit shown on the pricing page (Hobby plan: free platformcompute billed per second)

Use Cases

Custom model deployment

Fine-tuned model hosting

Testing GPU-intensive models

Scaling AI prototypes

Custom inference pipelines

Limitations & Considerations

$30 is one-time trial credits

Requires some DevOps knowledge

Cold starts for serverless models

GPU costs can vary significantly

Credit card required for continued use

Community Hub

Live

Join the discussion, share tips, and rate Cerebrium.

Quick Reactions

Add Discussion

Comments are moderated. Be helpful and respectful.

Recent Activity

0 comments

Ready to Get Started?

Join thousands of developers using Cerebrium

Start Building Now