Runpod Serverless
Plans
| Plan | Billed and limits | Price |
|---|---|---|
| Serverless GPU endpoint (16GB class) | $0.58/hr, metered per second · A4000, A4500, RTX 4000, RTX 2000 · 16 GB | Free |
| Serverless GPU endpoint (32GB RTX PRO 4500 Blackwell) | $1.15/hr, metered per second · RTX PRO 4500 Blackwell · 32 GB | Free |
| Serverless GPU endpoint (48GB L40/L40S/6000 Ada/MIG) | $1.75/hr, metered per second · L40, L40S, 6000 Ada, MIG 48GB · 48 GB | Free |
| Serverless GPU endpoint (96GB RTX 6000 Pro) | $3.49/hr, metered per second · RTX 6000 Pro · 96 GB | Free |
| Serverless GPU endpoint (24GB class) | $0.69/hr, metered per second · L4, A5000, 3090, MIG 24GB · 24 GB | Free |
| Serverless GPU endpoint (48GB A6000/A40) | $1.22/hr, metered per second · A6000, A40 · 48 GB | Free |
| Serverless GPU endpoint (80GB H100) | $4.79/hr, metered per second · H100 · 80 GB | Free |
| Serverless GPU endpoint (32GB 5090) | $1.58/hr, metered per second · 5090 · 32 GB | Free |
The first 8 of 10 plans, as runpod.io lists them, read 9 Oct 2026. The maker's own page has the final word. Every plan
Facts from Runpod Serverless's own pages
- Autoscaling
- The product overview says endpoints scale from zero to hundreds of workers with request demand, and flex workers scale to zero when idle. · runpod.io, 9 Oct 2026
- Billing
- Serverless compute is billed per second from worker start to full stop, rounded up to the nearest second. · runpod.io, 9 Oct 2026
- Cold starts
- FlashBoot is described as providing sub-200ms cold starts, while pre-warmed workers can provide zero cold starts. · runpod.io, 9 Oct 2026
- Compliance scope
- Runpod says compliance coverage can vary by workload, region, provider, and deployment model, and directs customers to confirm specific requirements with its team. · runpod.io, 9 Oct 2026
- Containers and runtimes
- Users can bring a Docker container and configure runtimes for languages and frameworks including Python, Node.js, Go, Rust, C++, PyTorch, TensorFlow, JAX, and ONNX. · runpod.io, 9 Oct 2026
- Deployments
- Workers can be deployed from GitHub, with code and Dockerfile built into a container image and deployed to an endpoint. · docs.runpod.io, 9 Oct 2026
- Founded
- 2022 · runpod.io, 28 Sept 2026
- Headquarters
- San Francisco, California, United States · runpod.io, 28 Sept 2026
- Integrations
- The documentation describes OpenAI-compatible endpoint integration with tools including n8n, CrewAI, and LangChain. · docs.runpod.io, 9 Oct 2026
- Intended workloads
- The product page identifies inference, fine-tuning, agent systems, rendering, and simulations as workloads for Serverless. · runpod.io, 9 Oct 2026
- Purpose
- Runpod Serverless runs containerized AI inference workloads behind API endpoints and scales workers based on demand. · runpod.io, 9 Oct 2026
- Regions
- The product page says users can deploy in global regions to place workloads closer to users. · runpod.io, 9 Oct 2026
- SDKs
- Runpod documents Serverless SDKs for Python, JavaScript, and Go. · docs.runpod.io, 9 Oct 2026
- Security and compliance
- Runpod states that it completed SOC 2 Type II and SOC 3 examinations, maintains HIPAA and GDPR programs, and holds an ISO/IEC 27001:2022 certification for its information security management system. · runpod.io, 9 Oct 2026
- Storage
- The product page describes persistent network storage that scales with workloads. · runpod.io, 9 Oct 2026
- Support
- Runpod documentation directs users to submit a support request, check the service status page, or join its Discord community. · docs.runpod.io, 9 Oct 2026
Where Runpod Serverless stands on Freedom251
- Model Hosting#16 of 25