About Runpod
Runpod is a cloud platform for renting GPUs and running AI workloads. It offers GPU pods, serverless endpoints and multi-node clusters across many regions.
What it does
You can launch a GPU pod for development, deploy an API endpoint that scales with incoming requests, or run a multi-node cluster for distributed training. A hub lets you deploy open-source models and ready-made templates. The control plane lets you manage your own compute and hosted options from one place.
Key features
- On-demand GPU pods in many regions
- Serverless GPU endpoints for API workloads
- Multi-node GPU clusters for training
- Hub with open-source models and templates
- Custom Docker templates for your own setup
Who it's for
- Developers who need GPU time on demand
- Teams serving models in real time
- Researchers running fine-tuning experiments
Pricing and good to know
You can get started for free and pay for the GPU time you use. Rates vary by GPU type and region, so compare them on the pricing page before a long run. Storage and network costs can apply, so read the billing terms before you start a large job.
Highlights
- Launch on-demand GPU pods
- Deploy serverless GPU endpoints
- Run multi-node clusters for training
- Start from hub models and templates
- Build custom Docker templates
