A real cluster in two minutes.
Yours from €5.21 a month.
No cloud account. Nothing to install. No card to start. A writable Kubernetes environment that is yours alone, with an app already answering over HTTPS by the time you get there - and when you decide to keep it, it costs less than the bare virtual machine you would have run it on.
- to ready in the recording
- 82sto ready in the recording
- to start, no card
- €0to start, no card
- a month to keep it
- €5.21a month to keep it
- cloud accounts to connect
- 0cloud accounts to connect
Plays from YouTube. Open on YouTube instead
Four steps, and you are the one typing
The recording above follows exactly this, start to finish, with nothing cut.
Start it
One click in the portal, under Clusters then Create cluster, or one command in your terminal. There is no cloud account to connect, nothing to install locally, and the free trial takes no card.
ankra cluster playground createWatch it come up
Four phases run in front of you: reserving capacity, starting the cluster, connecting the agent, deploying the starter stack. The recording clocks it at eighty-two seconds; one to three minutes is the honest range to plan around.
ankra cluster playground status <cluster-id>Find it already serving
A trial arrives with a starter app live on its own hostname over HTTPS, and a security baseline already scanning your images, before you have typed anything. The status command prints the URL. It is a real workload on a real cluster, not a screenshot of one.
curl -I https://podinfo.<cluster>.<org>.ankra.ccShip something of your own
Deploy a stack through the same pipeline your production deploys use, from the stack builder, the CLI, or by asking the AI assistant. What you learn here is not throwaway knowledge about a sandbox dialect.
ankra cluster select playground
ankra stack-profiles apply <profile> --deployPrefer to ask rather than type? Press Cmd+J in the portal and say "deploy nginx with 2 replicas" - it builds the stack and ships it through the same pipeline your production deploys would use.
Everything in it is the real thing
A sandbox that fakes the interesting parts teaches you nothing about the product. This one does not fake any of them.
Real nodes, real pods
It is a virtual cluster with its own API server, not a shared view of ours. Nodes report capacity, pods schedule and restart, events land where you expect them. Nothing is stubbed for the demo.
Real TLS, real hostname
Anything you publish with an Ingress gets a public HTTPS hostname under your playground's own domain, with a certificate that actually validates, so you can point a browser or a webhook at it.
Real security findings
A security baseline is installed with the environment, so the Security surfaces carry vulnerabilities scanned out of your own images rather than an empty state waiting for you to configure something.
Yours alone
The environment is writable and single-tenant. Deploy things, delete things, break things. It is not a shared read-only tour with the interesting buttons disabled.
The same control plane
Stacks, add-ons and the AI assistant behave exactly as they do on a cluster you own, because it is the same platform driving both. Ask it to deploy something and watch the pipeline run.
The same portable output
Everything you build is standard Helm charts and Kubernetes manifests. It moves to a cluster of your own unchanged, because it was never playground dialect to begin with.
The cheapest real Kubernetes cluster on the internet
A whole cluster, with storage, TLS, a public hostname and the control plane included, for less than most clouds charge for the control plane alone.
| Size | Yours to deploy into | Storage | Pods | Per month | Works out to |
|---|---|---|---|---|---|
| TrialFree | 1 vCPU / 2 GB | 20 GB | 30 | Free | One per organisation |
| Small | 2 vCPU / 4 GB | 50 GB | 40 | €5.21 | €2.61 per vCPU |
| Medium | 4 vCPU / 8 GB | 100 GB | 60 | €8.06 | €2.02 per vCPU |
| Large | 8 vCPU / 16 GB | 200 GB | 100 | €15.19 | €1.90 per vCPU |
| Extra large | 16 vCPU / 32 GB | 400 GB | 150 | €28.01 | €1.75 per vCPU |
The control plane is free
The API server, CoreDNS, the Ankra agent, the ingress controller and cert-manager run on top of the size you buy, not out of it. Roughly two vCPU and two gigabytes per environment, absorbed. Amazon charges USD 0.10 an hour, about USD 73 a month, for an EKS control plane on its own, before a single node.
Cheaper than the machine underneath it
Every paid size is priced below the bare virtual machine of the same shape at Europe's cheapest cloud. A virtual machine is not a cluster: you would still have to install Kubernetes on it, keep it patched, and wire up DNS and certificates yourself.
One line, nothing bolted on
Storage is bundled. There is no per-vCPU, per-GB, egress, IP or GPU charge on top, no load balancer to rent for a hostname, and a playground never counts toward your plan's worker-vCPU metering.
Billed by the second
Charging starts when the environment reports ready and stops when you destroy it, prorated across a resize. The price you order at is your price of record: later catalogue changes apply to new orders only, never to something you are already running.
Order, follow, resize and destroy from wherever you already are - the portal, the API, the CLI, or by asking Ankra AI. Anything that costs money names the size and the monthly price and asks you first.
ankra cluster playground plans
ankra cluster playground create --size small
ankra cluster playground resize playground --size mediumThe things you would otherwise find out later
A cheap environment is only cheap if nothing about it surprises you.
- How long does the free trial last?
- As long as somebody is around, plus a grace period once everyone goes quiet: three hours on the free plan, a week if your organisation is on a paid plan. Signing back in resets the clock, and you get an in-app and email warning before it goes. Paid sizes do not idle-expire at all: one stays until you destroy it.
- What does it cost to keep one?
- From €5.21 a month for Small, flat, storage included. Ordering a paid size needs a billing plan and a card on file; the free trial needs neither. Paid environments share an 8 GB memory budget per organisation, and the size picker offers a limit increase when you want more.
- Can I resize it later?
- Yes, in place, from Settings or the CLI. Only the quota and the price change, billed prorata across the change, and nothing you deployed is touched. The free trial has one fixed shape, so ordering a paid size is the upgrade path.
- How do I expose something?
- With an Ingress, which gets a hostname under your playground's domain and a certificate automatically. Not with a LoadBalancer Service: the playground host runs no load balancer implementation, so a LoadBalancer's external IP stays pending forever.
- What is not included?
- GitOps repositories and kubeconfig access, which are for clusters you own, and backups. Storage is local to the playground host and is not backed up, so keep anything you cannot lose in your own repositories and registries.
- Can I run production on it?
- No. Playgrounds are for development and evaluation, they carry no personal data, and they are excluded from the SLA. When something outgrows that, move it to a cluster of your own; the stacks come across unchanged.
Keep it, or graduate it
The playground used to be a demo you were expected to leave. Now it is also somewhere to stay.
A paid size is a permanent development cluster, and that is what most people end up using it for: one service and its dependencies on Small, a full stack with its datastore and observability on Medium, a shared team environment on Large. It does not expire, it resizes in place, and at these prices it is usually cheaper than the laptop time spent keeping a local cluster alive.
When something outgrows development and evaluation, the next step is a cluster of your own: import one you already run, or have Ankra build one on your account with a provider you choose. The stacks you wrote in the playground move across as they are - they are standard Helm charts and Kubernetes manifests, and they were never playground dialect to begin with.
From prompt to production.
No tickets in between.
Describe what you want to ship and let Ankra's AI generate, deploy, and operate it. Free forever for small teams.