Why we're not charging you yet

RunOS hit 1.0 on June 30, 2026. It's generally available, not a beta, not a preview. And it's free: unlimited nodes, unlimited clusters, no credit card. That's not an introductory offer with a countdown timer hidden in the terms. It's just where we are right now.

We could invent a price today. Pick a number per node, a number per service, round it to something that looks considered, and ship a pricing page that looks like everyone else's. We didn't, because we don't think that's how you should price infrastructure software, and we're not going to guess at a number just to have one.

So instead of a pricing tier chart, you're getting an argument. Read it, poke holes in it, run your own numbers through the calculator halfway down. Then start building if it holds up.

The markup you've been paying without noticing

If you've deployed on Heroku, Railway, Render, or Vercel, you've paid a markup on compute. Not a huge, obvious one. A quiet one, baked into the per-dyno or per-instance price before it ever reaches your invoice.

You don't see the gap because you're not meant to. The platform buys capacity from AWS or GCP at one rate, then resells it to you at another, and the difference funds the platform. That's a completely reasonable business, but it means the number on your bill was never the cost of your server. It was the cost of your server plus whatever the platform needed to add.

Most developers never check the underlying rate, because there's no reason to. The bill arrives pre-baked, round, and plausible. It looks like the cost of running your app. It's actually the cost of running your app inside someone else's margin.

See the gap yourself

We'd rather you check the math than take our word for it. Pick a node count and a CPU size below. It's the same cluster, priced across twelve providers. RunOS doesn't add anything on top of any of them: what you see for Hetzner, AWS, or DigitalOcean is what that provider charges, full stop.

What BYOC means in practice

RunOS is bring your own cloud. You keep your own account with Hetzner, AWS, or whoever you pick, and you pay them directly. RunOS never touches that invoice, never sits between you and the provider, and never takes a cut of it.

Underneath, RunOS runs standard Kubernetes. You get a real kubeconfig for your cluster, the same one you'd get from running Kubernetes yourself. That means nothing is hidden from you and nothing is locked to us: if you ever want to walk away from RunOS, your cluster keeps running and your kubeconfig still works. No lock-in, because there's no proprietary layer to be locked into.

What's actually free right now

Not a trimmed-down free tier designed to get you hooked before the good stuff shows up behind a paywall. All of it, today:

• All 20+ managed services: PostgreSQL, MySQL, ClickHouse, Valkey, Kafka, RabbitMQ, MinIO, Grafana, Prometheus, and more.
• All six ways to deploy, on any Ubuntu server, including boxes sitting behind NAT with zero inbound ports open.
• The AI stack: vLLM, Ollama, LiteLLM, and Langfuse, running on your own GPUs.
• Standard Kubernetes underneath the whole thing, no fork, no proprietary control plane.

No node cap. No cluster cap. No feature we've quietly fenced off for a plan that doesn't exist yet.

Where this goes next

We're honest that we don't have a pricing model announced yet, because we don't have one decided yet. We'd rather take the time to get it right than back into a number and spend a year apologizing for it.

Some things you might expect from a bigger deployment, like audit logs, SSO, and SOC 2 compliance, aren't built yet. They're on the roadmap, not in the product, and we're not going to pretend otherwise on a pricing page.

If we ever do introduce a paid tier, it'll be announced here first, with real numbers and real reasoning, not sprung on people already running RunOS in production. Nobody who's building on the free version today should worry about the rug getting pulled.