RunPod
Rents cloud GPUs on demand for AI work: dedicated GPU Pods, Serverless endpoints for inference and multi-GPU Clusters.
RunPod site ↗ Affiliate link Disclosure
Is it actually free? Paid
No free plan is listed on RunPod's pricing page. You pay as you go for GPU time and storage.
Pricing
Pay-as-you-go, billed per hour or per second. On RunPod's pricing page (updated Oct 8, 2026), Secure Cloud Pods range from $0.27/hr for an RTX A5000 to $8.99/hr for a B300; for example, an RTX 4090 is $0.89/hr and an H100 PCIe $2.89/hr. Community Cloud rates are lower. Storage starts at $0.05/GB a month. Reserved clusters and enterprise contracts are priced through sales.
What it offers
- Pods: dedicated GPU instances for development and long-running jobs, across 30+ regions.
- Serverless: GPU workers for API inference, billed on usage.
- Clusters: multi-GPU clusters of up to 64 GPUs with shared storage, no commitment.
- Public Endpoints: pre-deployed image, video, audio and language models you call by API.
Best for
Running or fine-tuning open models on a big GPU you don't own, paying only while it runs.
Not great for
People who want a ready-made chat app; you manage the software on the GPU yourself.