Now in Public Beta
Run AI workloads on GPUs.
70% cheaper than Modal.
Deploy containers, train models, and run inference. Multi-provider GPU cloud with per-second billing. No infrastructure to manage.
inference.py
import velar
app = velar.App("my-inference")
image = velar.Image.from_registry( "pytorch/pytorch:2.1.0-cuda12.1-cudnn8-runtime").pip_install("transformers", "accelerate")
@app.function(gpu="A100", image=image)def generate(prompt: str): from transformers import pipeline pipe = pipeline("text-generation", model="meta-llama/Llama-2-7b") return pipe(prompt, max_length=200)
# Deploy & runapp.deploy()result = generate.remote("What is machine learning?")Everything you need to run GPU workloads
A complete platform for deploying, scaling, and monitoring AI infrastructure.
Multi-Provider GPUs
Access A100, A10, L4 GPUs from RunPod & Vast.ai. Best price guaranteed.
Per-Second Billing
Pay only for what you use. No minimums, no commitments.
Python-Native
Define everything in code. No YAML, no Docker, no Kubernetes.
Sub-Minute Deploys
From code to running GPU in under 60 seconds.
Scale to Zero
Pay nothing when idle. Scale up instantly on demand.
Built-in Monitoring
Real-time logs, cost tracking, and deployment status.
Transparent GPU pricing
Simple per-second billing. No hidden fees, no long-term commitments.
| GPU | VRAM | Velar | Modal | Savings |
|---|---|---|---|---|
| A100 80GB | 80 GB | $3.20/hr | $4.00/hr | 20% cheaper |
| A10 24GB | 24 GB | $1.40/hr | $1.10/hr | — |
| L4 24GB | 24 GB | $0.85/hr | $0.80/hr | — |
Start building with $10 free credits
No credit card required. Deploy your first GPU workload in minutes.
No credit card required