Grafana Cloud Pricing Per User: What an Active User Costs

Grafana Cloud charges $8.00 a month per active user above three included users, falling to $7.25 for users 26 to 50 and $6.80 for 51 and above. Ten people cost $56 a month, twenty five cost $176, and a hundred cost $700.85, all on top of the $19 platform fee and whatever your metrics, logs, traces and synthetic checks meter. The word doing the work in that sentence is active: Grafana bills the people who actually opened it that month, not the seats you provisioned, which is a more forgiving model than most of this market ships.

Almost every write-up of this number gets it wrong in one of two ways. Some quote a flat $8 and stop, missing that there is a published volume ladder underneath it. Others quote $20 per user, which is not a figure that appears anywhere on Grafana's pricing page today. Everything below was read on 15 September 2026 from grafana.com/pricing, in US dollars, parsed from the users block on that page rather than from a directory listing.

How much does Grafana Cloud cost per user?

Three published bands, three included users, and a platform fee sitting underneath the whole thing.

Grafana describes these as automatic volume discounts, which means you do not negotiate them and you do not ask for them. It does not say whether a band rate applies only to the users inside that band or re-prices everybody once you cross the line. The figures in this article apply them marginally, to the count of billable users, because that is the more conservative reading and it matches how the rest of this market publishes ladders. If Grafana applies them flat, real invoices at high headcount are lower than what you see below.

What a team actually pays

The three included users are the part people forget, and they matter most at small headcount because they are a third of a nine person team and a rounding error on a hundred.

Read the right hand column and you find something slightly odd. The effective cost per person does not fall steadily as you grow, which is what a volume discount is supposed to do. It climbs from $3.20 at five people to a peak of $7.19 somewhere around fifty, and only then starts coming down, reaching $6.88 at 250. The three free users are responsible for the whole shape: they subsidise a small team heavily and a large one barely at all. So the growth stage that feels most expensive per head is the one in the middle, roughly 25 to 60 people, where the freebie has stopped mattering and the discount has not yet arrived.

What counts as an active user?

An active user is someone who actually used Grafana in the billing month, not someone who has an account. Grafana counts use broadly, including viewing dashboards, receiving and acting on alerts through Slack or Microsoft Teams, the mobile and desktop apps, and the command line tool. Provisioning a login for somebody who never signs in does not produce a charge.

That distinction is worth more than it looks. Most per-seat vendors bill the account, which means every departed employee and every just-in-case login is money until somebody remembers to clean up the user list. Grafana's meter cleans up after itself. If your support team logs in during an incident in March and not again until July, you pay for March and July. Credit where it is due: this is the friendlier end of per-user pricing and it is not how most of this category works.

The seat is not the only meter, and it is rarely the biggest

Per-user is one line on a Grafana Cloud invoice and it is usually not the one that surprises anybody. The platform fee is $19 a month on Pro. Metrics run $6.50 per 1,000 active series beyond the included 10,000. Logs and traces bill in three parts, for processing, writing and retention. And synthetic checks meter by test execution, which is where the arithmetic gets genuinely counterintuitive, because probes, check count, run time and interval all multiply together rather than adding up.

That last one deserves a number, because it is the line most teams underestimate by an order of magnitude. A single HTTP check running every minute from five probe locations consumes 216,000 test executions a month, which is more than double the free allowance, and costs about $77 a month on its own. We worked through the whole meter, the rate ladder and the round-up rule that doubles a check's cost when it slows past sixty seconds, in the Grafana Cloud synthetic monitoring pricing breakdown. Set next to a $56 user bill for ten people, the checks are the expensive half.

The broader pattern is that Grafana Cloud has a lot of independent meters and each one is individually reasonable. The bill is the product of decisions nobody experiences as a purchase: adding a region for reliability, tightening an interval after an outage, turning on a new log source. If you run a consumption platform of any kind it is worth having something that tells you when the curve bends, because the invoice arrives a month after the decision that caused it, and a real-time budget alert is a cheaper way to find out than the finance review is.

Who on your team actually needs a Grafana seat?

This is the question that decides the bill, and it is worth asking properly rather than defaulting to giving everybody access.

Engineers building and debugging services need the real thing. They need to pivot from a latency spike to the trace to the log line, and no summary view substitutes for that. Buy them seats and do not think about it again.

The larger group is different. Support leads, account managers, customer success, the people who run status communications during an incident, and executives who want to know whether last quarter had an outage. What they need is a far narrower question answered: is it up, was it up, and what do we tell customers. They are not going to build a panel. At $8.00 each, twelve of those people is $96 a month and $1,152 a year, for a product most of them will open during incidents and ignore the rest of the time.

Pointing that group at a status page and an uptime tool instead is usually both cheaper and more useful, because the thing they actually want is a public page they can link a customer to. Our own plans include unlimited team and client seats on every tier rather than charging per person, and branded status pages come with all of them. A team of twenty five split the sensible way, engineers in Grafana and everybody else on a status page, pays for a handful of Grafana users rather than twenty five.

Grafana Cloud per-user pricing compared with the rest of the market

Per-user pricing in monitoring and on-call runs from free to genuinely expensive, and Grafana sits at the cheap end.

Two things stand out. Grafana is roughly a third of what the incident response tools charge for a person, because it is selling dashboards rather than on-call rotations and paging. And it is the only vendor in that table whose per-user rate goes down as you add people rather than up: Hyperping's seat rate rises from $7 to $12 as you climb its tiers, which we set out in the Hyperping pricing breakdown, and PagerDuty publishes no volume ladder at all.

The comparison that matters more than any of these, though, is against the checks themselves. If you are evaluating Grafana Cloud because you want to know when your site goes down, the user line is the small half of that decision. The metered half is in the synthetic monitoring pricing comparison, which prices Datadog, New Relic, Dynatrace, Grafana, AWS and Checkly per check run, because every one of them meters a different unit and the unit decides the bill.

Does the free plan include users?

Yes, three of them, and the Pro plan includes the same three. This is a pattern worth noticing on Grafana's pricing page, because it repeats: the $19 Pro platform fee includes exactly the same synthetic monitoring allowance as the free plan too, 100,000 API and 10,000 browser test executions. What the platform fee buys is 13 months of retention instead of 14 days, email support, and the right to exceed the caps and pay for the overage. It does not buy included capacity.

So the honest summary for a small team is that Grafana Cloud is free, genuinely free rather than trial free, until you need a fourth person looking at dashboards or a fourth check running frequently. After that, both meters start at once.

What we would do

Count the people who will build things in Grafana and buy exactly those seats, because at $8.00 with a usage-based definition of active, they are good value and the alternative is engineers without observability. Then count the people who only need to know whether the site is up, and do not buy them anything. Give them a status page and an alert channel. That split usually turns a twenty five person Grafana bill into a six person one, and it produces a better experience for the nineteen, because a dashboard built for a distributed tracing workflow is not what somebody wants at eight in the morning when a customer is asking whether the API is down.

If you want the other half of the Grafana Cloud bill worked out properly, the execution meter, the rate ladder, the sixty second round-up cliff and where the free allowance actually runs out, it is all in the Grafana Cloud synthetic monitoring pricing breakdown, with a calculator that runs Grafana's own published formula over your configuration.