Grafana Cloud costs $0 on the free tier, starts at $19 a month on Pro plus usage, and jumps to a $25,000 a year spend commitment on Enterprise. The plan fee is almost never what you actually pay. Grafana Cloud is not one bundled price, it is roughly a dozen separate meters, and the two that surprise people most are logs and synthetics. Logs are advertised from $0.050 per GB, but that is only the process component: add write at $0.400 and retain at $0.100 and the real number is $0.550 per GB, eleven times the figure most write ups quote. Synthetic checks are billed per probe location per minute, so a single URL checked every minute from three probes is $64.80 a month by Grafana's own worked example.
Every rate below is the US dollar list price published on Grafana's own pricing page, read in September 2026. The arithmetic is ours and each figure is shown so you can check it against your own usage.
Grafana Cloud pricing 2026: the three tiers
The tier decides your support level, your retention and your discount structure. It does not cap your usage, which is the single most important thing to understand before you model a budget. Pro has no ceiling. You can run a $4,000 month on a plan whose headline price is $19.
| Tier | Base price | Usage above free tier | Metrics retention | Logs, traces, profiles retention | Support |
|---|---|---|---|---|---|
| Free | $0, always free | Not available, usage is capped at the allowance | 14 days | 14 days | Community |
| Pro | From $19 / month | Pay as you go with volume discounts | 13 months | 30 days | 8x5 email |
| Enterprise | From $25,000 / year commit | Committed rates, custom quote | Custom | Custom | Enterprise support |
Put those two paid numbers next to each other and the gap is the story. Pro's base is $228 a year. The Enterprise entry commitment is $25,000 a year. That is 110 times the Pro base fee, and there is nothing in between: no mid tier, no published $5,000 contract. If you need Enterprise features such as committed rates, custom retention or a signed support SLA, the smallest cheque is a five figure annual commitment.
What the Grafana Cloud free tier includes
The free tier is genuinely generous by observability standards and it is the reason Grafana Cloud shows up on so many shortlists. It is also where the first real budgeting mistake gets made, because the allowances are not the same shape as the things you will eventually buy.
| Signal | Free allowance per month | Rate once you exceed it |
|---|---|---|
| Metrics | 10,000 active series | From $6.50 / 1,000 series |
| Logs | 50 GB ingested | From $0.550 / GB, all components |
| Traces | 50 GB ingested | From $0.550 / GB, all components |
| Profiles | 50 GB ingested | From $0.550 / GB, all components |
| k6 load testing | 500 virtual user hours | From $0.150 / virtual user hour |
| Grafana users | 3 active users | From $8.00 / active user |
| IRM users | 3 active IRM users | From $20.00 / active IRM user |
| Synthetic monitoring | Not listed as a free allowance | From $5.00 / 10,000 API test executions |
That last row is worth pausing on. Grafana's free tier table names allowances for metrics, logs, traces, profiles, k6 and users. Synthetic monitoring executions are not among them. If uptime checking is the job you came to Grafana Cloud to do, the free tier is not the plan you are evaluating, and the check volume you need is metered from the first execution.
The user allowance catches teams out too. Three active users is a founding team, not a company. The fourth person to open a dashboard costs $8 a month, so a ten person team is seven paid seats at $56 a month, or $672 a year, on top of every data meter. Add on call and it gets steeper: IRM seats are $20 each, so ten people carrying a pager is $140 a month, $1,680 a year, before a single metric is ingested.
How much does Grafana Cloud cost per meter
This is the full published rate card. Note the structure carefully, because it is the thing that makes Grafana Cloud hard to forecast: logs, traces and profiles are each billed in three components rather than one, and several products switch billing unit entirely.
| Product | Billing unit | List rate |
|---|---|---|
| Metrics | Active series | From $6.50 / 1,000 series |
| Logs | GB, three components | $0.050 process + $0.400 write + $0.100 retain |
| Traces | GB, three components | $0.050 process + $0.400 write + $0.100 retain |
| Profiles | GB, three components | $0.050 process + $0.400 write + $0.100 retain |
| k6 load testing | Virtual user hour | From $0.150 / VUh |
| Synthetics, API | Test execution | From $5.00 / 10,000 executions |
| Synthetics, browser | Test execution | From $50.00 / 10,000 executions |
| Frontend observability | Session | From $0.750 / 1,000 sessions |
| Grafana visualization | Active user | From $8.00 / active user |
| IRM, on call and incident | Active IRM user | From $20.00 / active IRM user |
| Kubernetes monitoring | Host hour | $0.0100 / host hour |
| Application observability | Host hour | $0.025 / host hour |
| Database observability | Host hour | $0.070 / host hour |
The log pricing trap: $0.05 per GB is really $0.55 per GB
Search for Grafana Cloud log pricing and you will repeatedly see the figure $0.05 per GB. It is a real published rate and it is also the smallest of the three components you are billed for. Grafana lists process at $0.050, write at $0.400 and retain at $0.100. A gigabyte of logs that you ingest, store and keep touches all three, so the number to budget with is $0.550 per GB. Quoting the process rate alone understates the bill by a factor of eleven.
On a terabyte a month, which is not a large logging footprint for a mid sized production estate, that is the difference between an imagined $48.70 and an actual $535.70. The same three component structure applies to traces and to profiles, so the error compounds across every signal you turn on.
This is also why a Grafana Cloud bill can move sharply without anybody deploying anything. Log volume is a downstream effect of what your services emit, and a single chatty release, a retry storm or a new debug logger can multiply ingest overnight. Treating an unexplained jump in ingested gigabytes as a volume anomaly worth alerting on in its own right is the cheapest control available here, because by the time it shows up on the invoice you have already paid for a full month of it.
What Grafana Cloud synthetic monitoring costs per URL
Grafana Cloud sells synthetic monitoring, powered by k6, with API endpoint, multi step, scripted and browser checks run from public or private probe locations. If you are pricing Grafana Cloud as an uptime monitoring tool, this is the meter that matters, and it has one property that changes the arithmetic completely.
Grafana defines a test execution as a synthetic test running in a probe location, per minute of run time, and states plainly that executions are counted in both public and private probe locations and billed the same way. Probes multiply. Grafana's own documentation works the example: a check running every minute across three probes is 129,600 test executions a month. At $5.00 per 10,000 API executions that is $64.80 a month. For one URL.
Running the same volume as a browser check costs exactly ten times more, because browser executions are $50.00 per 10,000 against $5.00 for API. That is $648.00 a month for one URL.
| Configuration, one URL | Executions per month | API check cost | Browser check cost |
|---|---|---|---|
| Every minute, 1 probe | 43,833 | $21.92 | $219.16 |
| Every minute, 3 probes | 131,499 | $65.75 | $657.50 |
| Every 5 minutes, 1 probe | 8,767 | $4.38 | $43.83 |
| Every 5 minutes, 3 probes | 26,300 | $13.15 | $131.50 |
| Every 10 minutes, 1 probe | 4,383 | $2.19 | $21.92 |
| Every 15 minutes, 1 probe | 2,922 | $1.46 | $14.61 |
Three probes is not an extravagant setting, it is the responsible one. A single vantage point cannot tell a dead service apart from a bad route between one probe and your origin, which is why confirming a failure from more than one location before you page anyone is standard practice. Grafana charges you for that correctness in a straight line.
Here is the honest counterpoint, because the meter genuinely wins at the low end. At a 15 minute interval from one probe, $19 buys 13 URLs of Grafana synthetic checking. At 10 minutes from one probe it buys 8.7. If you have a handful of endpoints and you are content with a quarter hour of blind time, the metered rate beats any flat plan including ours. The direction reverses hard as soon as you want minute level resolution or multi probe confirmation: at one minute across three probes, $19 buys 0.3 of a URL.
Three worked Grafana Cloud bills
Rate cards are abstract, so here are three realistic configurations priced end to end at list.
| Line | Small startup | Growing SaaS |
|---|---|---|
| Pro base | $19.00 | $19.00 |
| Metrics | 25k series: $97.50 | 120k series: $715.00 |
| Logs | 100 GB: $27.50 | 400 GB: $192.50 |
| Synthetics | 5 API checks, 5 min, 1 probe: $21.92 | 20 API checks, 1 min, 3 probes: $1,314.99 |
| Extra Grafana seats | None, 3 users | 5 extra seats: $40.00 |
| IRM seats | None | 2 extra seats: $40.00 |
| Monthly total | $165.92 | $2,321.49 |
| Annual total | $1,991.04 | $27,857.88 |
The growing SaaS column is the one worth staring at. Its annual run rate has quietly passed the $25,000 Enterprise commitment threshold, which means that team is paying Pro list prices for spend that would qualify for committed Enterprise rates. If your Pro bill is trending past roughly $2,100 a month, the negotiation is worth having.
Note also what dominates that bill. Synthetics alone is $1,314.99 a month, 57 percent of the total, for 20 monitored URLs. That is 22 times what the same 20 URLs cost inside a flat monitoring plan, and it is the clearest case for keeping uptime checking on a separate, unmetered tool rather than running it through an observability meter.
Is Grafana Cloud expensive?
It depends entirely on which meters you turn on. As a metrics and dashboards platform for a small team it is competitive and the free tier is real. As a full observability stack at volume it is priced like every other enterprise observability platform, and the per component log billing makes it easy to underestimate by an order of magnitude.
As an uptime monitoring tool specifically, it is expensive, and not because Grafana is overcharging. The product is built for engineers who already live in Grafana and want their synthetic results in the same Prometheus and Loki stores as everything else. That integration is genuinely valuable and you pay for it per execution per probe. If what you need is to know within seconds that a URL, API or port stopped answering, and to have that reach a human by SMS, you are buying a very expensive version of a much simpler product.
How much does Grafana Cloud cost compared to a dedicated uptime monitor?
The comparison only makes sense on the synthetics meter, since a dedicated uptime tool does not ingest your metrics or logs. On that meter the gap is wide. Twenty URLs checked every minute from three probes is $1,314.99 a month on Grafana Cloud list rates. The same twenty endpoints sit inside a flat monitoring plan at $19 a month, or $59 with 30 second checks and room for 100 monitors.
That does not make Grafana Cloud the wrong choice. It makes it the wrong single choice. Most teams that run both keep Grafana for correlation and root cause, and keep a separate external checker for the detect and page loop, precisely because an external checker that lives outside your observability stack keeps working when your observability stack is the thing that broke. We wrote about how to draw that line in uptime monitoring to pair with APM, and the same reasoning underpins our uptime monitoring pricing page.
Does Grafana Cloud have a free tier for uptime monitoring?
Not a named one. Grafana Cloud's published free tier allowances cover metrics, logs, traces, profiles, k6 virtual user hours and user seats. Synthetic monitoring test executions are not listed among them, so plan on the $5.00 per 10,000 API executions rate applying to your check volume. Small check volumes at long intervals are cheap in absolute terms, but they are metered rather than free.
Why did my Grafana Cloud bill go up?
Four causes account for most unexplained increases. Metric cardinality grew, because a new label with many values multiplies active series and series are billed at $6.50 per thousand. Log volume grew, at $0.550 per GB all in. Someone added probe locations to existing synthetic checks, which multiplies executions by the number of probes with no change to the check count. Or headcount grew past three active users at $8 each and three IRM users at $20 each. None of those four require a deliberate purchase decision, which is exactly why they are the usual suspects.
What is the minimum spend for Grafana Cloud Enterprise?
Grafana publishes Enterprise as starting at a $25,000 per year spend commitment. There is no smaller published Enterprise contract and no published tier between Pro and Enterprise, so a team that needs committed rates, custom retention or enterprise support is looking at a five figure annual commitment as the entry point.
How Grafana Cloud compares with the other per-run meters
Grafana is not alone in billing synthetic checks by execution, and the same surprise arrives from different directions across the category. Checkly bills per check run with a separate, higher rate if you exceed the allowance rather than prepaying, which we broke down in Checkly pricing, and every vendor's rate is set side by side on our synthetic monitoring tools page.
The short version
Grafana Cloud's headline prices are honest and its free tier is real. The forecasting risk is structural rather than deceptive: a dozen independent meters, three component billing on your highest volume signals, and a synthetics meter that multiplies by probe count. Model your own series count, your ingested gigabytes and your check volume against the rates above before you commit, and price the uptime detection job separately, because that is the line item where the metered model costs the most and delivers the least differentiated value.