Grafana Cloud Review

Grafana Labs built the dashboard half the industry stares at, then wrapped Loki, Mimir, Tempo and Pyroscope around it and sold the bundle as a hosted service.

Grafana Cloud is that bundle: the open-source stack, operated for you, metered on data rather than on hosts or seats. My view after pricing it against the suites: it is the best default for a team that knows OpenTelemetry and Prometheus and wants to keep the option of self-hosting, and a worse fit for a team that wants one agent, one product and no assembly.

This review is for developers and engineering leads deciding between Grafana Cloud, Datadog, New Relic and the OpenTelemetry-native challengers.

It covers what the platform includes in 2026, how the series and gigabyte meters work, what Free, Pro and Enterprise change, and the two traps buyers miss: cardinality on the metrics meter and the three-part log rate. The five alternatives at the end carry prices from their own pricing pages.

Grafana Cloud is a hosted observability platform from Grafana Labs, the company founded in December 2013 around Torkel Odegaard's open-source dashboard project and now led by CEO Raj Dutt.

Grafana Labs says it is trusted by more than 10,000 customers and maintains Grafana, Loki, Mimir, Tempo, k6, Pyroscope, Faro, Beyla and Alloy, which are the components under the service.

The product set covers metrics (Prometheus-compatible, backed by Mimir), logs (Loki), traces (Tempo), profiles (Pyroscope), and a layer of packaged apps on top: Application Observability for APM-style service views, Kubernetes Monitoring, Frontend Observability, Database Observability, Synthetic Monitoring for API and browser checks, k6 performance testing and IRM for on-call and incident response.

Grafana Assistant and Investigations add an AI layer billed per active user and per million tokens.

Everything is queried in the Grafana UI with PromQL, LogQL and TraceQL, which is the point: a team that has learned the open-source tools loses nothing by moving to the cloud and nothing by moving back. Compared with Datadog and New Relic, it is the less integrated suite; compared with SigNoz and Better Stack, it is the broader one.

You send data with Grafana Alloy, the OpenTelemetry collector or any Prometheus remote-write client, and each signal lands on its own meter. Metrics are billed on active series, where an active series is a time series that received a data point in the previous 20 minutes.

Grafana Cloud bills the 95th percentile of active series over the month, which forgives roughly the top 36 hours of a 720-hour month, so a short spike does not set the invoice. Pro includes 1 data point per minute per series; send more and the bill is based on total data points instead.

Logs are billed in parts. Gigabytes processed are charged when the data arrives, gigabytes written are charged after optimisation with a 50 GB free allotment, and gigabytes retained are charged per 30-day increment beyond the included retention, which is 14 days on Free and 30 days on paid accounts.

Queries are free up to 100 times the written volume each month. Traces and profiles follow the same three-part rate.

The platform fee is the odd piece. Pro charges a flat monthly fee before any usage, and that fee buys the longer retention (13 months for metrics, 30 days for logs), the 8x5 email support and the ability to exceed the free allowances at all.

Free accounts hit a hard ceiling; Pro accounts bill for the overage. Enterprise is a custom contract with a high yearly minimum and metrics as low as $3 per 1,000 series, which only pays off well past a million series.

Every Grafana Cloud rate on this page was read on Grafana Cloud pricing in October 2026 and the billing mechanics on Grafana's invoice documentation the same month. Competitor figures come from Datadog pricing, New Relic pricing, SigNoz pricing, Better Stack pricing and Honeycomb pricing, read the same month.

The metrics meter is the one to watch. Pro includes the same 10,000 series as Free and charges per 1,000 above that, so 50,000 series cost $260 a month before any logs. A single label with high cardinality, a user ID or a request ID, can push a small service from hundreds of series to tens of thousands with no change in traffic. Logs and traces are cheaper than they look. The headline rates are $0.05 to process and $0.40 to write per GB, so a gigabyte above the 50 GB allowance costs $0.45 in total at the included 30-day retention; the $0.10 retain charge applies only to each extra 30 days kept, and queries up to 100 times the written volume are free. Visualisation users cost $8 each above three, which matters for a company where everyone opens dashboards, and users with enterprise plugins cost $55.

Cost scenario, computed from those rates: a five-engineer startup on ten hosts sending 100 GB of logs and 50 GB of traces a month, with all five using dashboards and metrics inside the included 10,000 series. Pro comes to the platform fee, $22.50 for the 50 GB of logs above the allowance, nothing for traces inside their allowance, and $16 for the two users above the free three: about $57.50 a month. Add Application Observability on all ten hosts and the bill rises by roughly $180 to about $238. The same team pays $129 on New Relic (50 GB above its free 100 GB plus two Standard full-platform seats) and $470 on Datadog (Infrastructure Pro at $15 per host, APM at $31 per host and per-GB log ingest) before any log is indexed.

Devshot data: Grafana Cloud is listed as an alternative to 11 other tools on Toolradar, the software directory we run, and its entry paid plan ranks joint 6th of 23 for price among the log management tools the directory prices, as of October 2026. Across the 210 monitoring tools Toolradar tracks, 64% offer a free tier; Grafana Cloud's is the one with the widest allowances among the commercial suites.

Grafana Cloud fits teams that already speak Prometheus and OpenTelemetry and want the stack operated for them. A platform engineer who has run Loki once and never wants to again is the ideal buyer: the queries and dashboards transfer, the data meters reward sampling discipline, and self-hosting stays a real option.

It also fits early-stage startups, because the free tier carries a live product and Pro's bill tracks data rather than headcount.

It is a weaker fit for teams that want one agent and one integrated product. Correlating a trace with its logs and the host it ran on works, but it takes configuration that Datadog and New Relic do for you.

It is also the wrong choice for a company that expects to pass a million metric series without an Enterprise contract, and for a team whose first need is external uptime checks and paging, where Better Stack bundles both for less.