Kubecost Pricing: What Kubecost Free, Enterprise Self-hosted and Enterprise Cloud Actually Cost

Kubecost does not publish a price for any paid tier. What it does publish, as of September 2026, is a permanently free Foundations plan covering unlimited clusters up to 250 cores combined, with unlimited users and 15-day metric retention. Above that ceiling you are into Enterprise Self-hosted, bought through the IBM store, or Enterprise Cloud, which is contact-only. Neither shows a number. The per-node and per-month figures circulating in search results are not on the vendor's pricing page, and you should not build a budget on them.

There is one more thing worth knowing before you start an evaluation: kubecost.com/pricing now redirects to Apptio's pricing page. IBM acquired Kubecost in September 2024 and has folded it into the Apptio product line, which is why the buying motion increasingly looks like an IBM enterprise purchase rather than a developer tool signup. Everything below was read off that page on 4 September 2026.

What are the Kubecost pricing tiers?

Three tiers, one of which has a real published limit and two of which do not have a published price.

The Enterprise Self-hosted listing carries a footnote that reads "For deployments above 3,200 VPCs, please contact us." Given the rest of the page is denominated in cores, it is worth asking your account team directly whether that ceiling is measured in VPCs or vCPUs, because the two produce wildly different answers about whether you are near it. That is a two-minute email that has saved buyers a surprise on renewal.

How much does Kubecost cost per node?

Kubecost does not publish a per-node price. Search results and third-party comparison pages quote figures in the region of $8 per node per month, a $449 monthly business tier, and various AWS Marketplace annual contract numbers. None of those appear on Kubecost's own pricing page today. Some were accurate for older packaging, some come from aggregator sites that infer rather than verify, and repeating them in a budget request is how a finance conversation goes wrong three months later.

The reason this matters more here than elsewhere is the acquisition. Product packaging and price lists change after an acquisition, and Kubecost has been through one. Treat any Kubecost price you did not read on an IBM or Apptio page, or receive in a written quote, as unverified. If you need a number for a budget cycle, the only reliable route is to request one, and to budget four to six weeks of enterprise sales cycle for it.

Is Kubecost free, and what is the catch?

Foundations is genuinely free rather than a countdown trial, and it is more generous than most free tiers in cloud cost management. Unlimited clusters, unlimited users and accurate allocation by namespace, deployment, label and pod is a real product, not a demo. The catch is two numbers.

The first is the 250-core ceiling, which is combined across clusters rather than per cluster. That sounds large until you count. A modest production estate of three clusters running 16-core nodes crosses 250 cores at roughly sixteen nodes total, which many teams pass in their first year. The second, and the one that bites sooner in practice, is the 15-day metric retention. Fifteen days is not long enough to see a monthly billing cycle, so you cannot answer month-over-month questions, you cannot show a quarter of trend to a finance team, and you cannot do the before-and-after comparison that justifies an optimization project. Retention, not cores, is usually what pushes teams off the free tier.

Kubecost vs OpenCost: does the free tier make Kubecost redundant?

No, and the distinction is worth being precise about because the two names get used interchangeably. OpenCost is the CNCF open-source project that provides the core cost allocation model. Kubecost is the commercial product built on it, adding the interface, longer retention, alerting, governance features and support. If you have platform engineering capacity and want to run and maintain the allocation layer yourself, OpenCost is a legitimate answer and costs nothing but time. If you want the product experience, Kubecost Foundations gives you a large slice of it for free. We have compared the two in detail in our Kubecost vs OpenCost breakdown.

What does Kubecost not do at any price?

Kubecost is a Kubernetes cost allocation and monitoring tool. It tells you what a namespace, deployment, label or team costs, and it surfaces rightsizing recommendations. It does not run your cluster. It will not resize your pods automatically, manage node provisioning, handle spot capacity or act on your commitments, and no tier changes that. Teams who want the bill to go down without a human in the loop are shopping in a different category, which is where automated pod rightsizing platforms such as StormForge, CAST AI, Zesty and Sedai live. Our wider Kubernetes rightsizing tools comparison covers who changes what.

The second boundary is scope. Kubecost is Kubernetes-shaped, so it sees your clusters clearly and the rest of your cloud bill much less clearly. If your CFO is asking what the whole estate costs per product line, including the managed databases, the load balancers, the object storage and the SaaS licences that sit next to the clusters, a Kubernetes tool answers a fraction of the question. That is a common reason teams end up running Kubecost alongside a broader allocation platform rather than instead of one.

It is also worth being clear that cost monitoring is not availability monitoring. Kubecost will tell you that a service got more expensive; it will not tell you that its public endpoint stopped answering. If nobody currently owns that second question, a straightforward uptime monitor that checks your endpoints and ports every 30 seconds is a cheaper fix than trying to make a cost tool do it, and the two datasets read well side by side when you are working out whether a spike was traffic or an incident.

When should you move off the Kubecost free tier?

There are three honest triggers, and none of them is simply crossing 250 cores.

The first is when someone outside the platform team starts asking for the numbers. The moment finance wants a monthly report, 15-day retention is disqualifying and no amount of exporting to spreadsheets fixes it sustainably. The second is when you need to prove a saving. Optimization work is funded on before-and-after evidence, and you cannot produce that with a two-week window. The third is governance: budgets, alerting thresholds and access control for multiple teams are the features that stop cost management being one engineer's side project.

If none of those three apply yet, staying on Foundations is a perfectly good decision and you should spend the budget on something that is currently hurting. If one of them does apply, the useful move is to price the alternatives at the same time rather than defaulting to the upgrade, because the thing you now need, which is retention plus reporting plus multi-team governance, is exactly what a general cloud cost platform does, and several of them publish prices. Our Kubecost alternatives page covers the like-for-like Kubernetes options, and the Kubernetes cost management tools comparison covers the wider field.

How should you evaluate Kubecost pricing against alternatives?

Compare on the two axes that actually differ, rather than on feature checklists that all look the same. The first is scope: does the tool see only Kubernetes, or the whole bill? Kubernetes-only tools are cheaper and sharper, and they leave a gap the moment your reporting is by product rather than by cluster. The second is whether the price is knowable in advance. A published price lets you decide in an afternoon. A quote-only price costs you a sales cycle before you can even compare, which is a real cost when you are running a bake-off between three vendors.

Concretely, work out your combined core count and your reporting horizon first, because those two numbers decide whether this is a free-tier conversation at all. Then get quotes in parallel rather than sequentially. Then compare the total, including the internal time each option costs you: self-hosting OpenCost is free in licence terms and not free in engineering terms, and an enterprise purchase is expensive in licence terms and cheap in engineering terms.

If what you need is allocation and forecasting across the whole estate rather than just the clusters, Costanalyst pricing is published at $99, $299 and $799 per month, connects read-only with no controller in your cluster, and covers cloud and SaaS spend together. That is a different product from Kubecost rather than a straight swap, and the honest test is which question your organization is actually being asked: what does this namespace cost, or what does this product line cost.