Datadog vs New Relic vs Grafana Cloud: which pricing model wins

The question people ask is which one is cheaper. There is no single answer, because the three do not bill on the same axis. Datadog meters per host, New Relic meters per person and per GB of data, and Grafana Cloud meters per usage primitive. So the same monitoring workload can be cheapest on a different tool depending on whether hosts, engineers or data volume is your biggest number. Pick the tool whose meter matches the number you can keep small.

Three different meters, not three prices

• Datadog: per host, then a separate meter stacked per product (infrastructure, APM, logs), and logs billed twice, once to ingest and once to index.
• New Relic: per full platform user, plus per GB of data ingested, with the first 100 GB a month free.
• Grafana Cloud: per usage primitive, mostly metric series and log GB, with users mostly free.

Datadog: per host, and the meters stack

Datadog prices on your host count and then adds a separate line per product: about 15 dollars per host a month for Infrastructure on the annual Pro plan, about 31 dollars per host for APM on the hosts you trace, and logs billed twice, roughly 0.10 dollars per GB to ingest and again per million events to index. There is no free production tier, so you are on the meter from the first host. The model is predictable when hosts are your main number, but the extra meters, especially logs, are where the bill quietly climbs. If Datadog is already your platform, the sibling guide on why a Datadog bill keeps climbing covers which line to cut first.

New Relic: you pay for people and data, not hosts

New Relic ignores host count entirely and bills on two things: how many people need full access, and how much data you send. The free tier includes 100 GB of ingest a month and one full platform user. On the Standard plan the first full user is about 10 dollars, each additional full user is about 99 dollars a month up to five users, and lighter core users are about 49 dollars; data past the free 100 GB is about 0.40 dollars per GB, or about 0.60 dollars per GB on Data Plus. So the swing cost is how many engineers need a full seat, not how many servers you run. It is cheap when a small group owns observability across a large fleet, and it gets expensive when the whole team needs full access.

Grafana Cloud: usage-based, with the most generous free tier

Grafana Cloud bills on the raw usage primitives and keeps users mostly free. The free tier is the most generous of the three: about 10,000 metric series, 50 GB of logs and 50 GB of traces, three users, at 14-day retention. On Pro you pay a 19 dollar a month platform fee, about 6.50 dollars per 1,000 active metric series, about 0.40 dollars per GB to write logs (plus small processing and retention charges), and 8 dollars per active user beyond the three free ones. Cardinality is the bill here: high-cardinality labels multiply your series count, so the same dashboards can cost very different amounts. It gives the lowest floor and the most control, in exchange for being the most hands-on to run.

The rule of thumb

The cheapest tool is the one whose meter tracks the number you can keep smallest. Many hosts but a small team: Datadog is predictable, just watch the log line. A big team where everyone needs full access but a modest fleet: Datadog per-host can flatter you, but check the full-user count against New Relic before you assume it wins. Data-heavy, cost-sensitive and willing to self-serve: Grafana Cloud has the lowest floor and the best free tier. None of that is a substitute for adding up your own meters, which is what the estimator is for. (List prices, checked July 2026, they move.)

When logs are the line that runs away

If the number that keeps growing is log volume rather than hosts or users, the split-meter model is the problem, not the vendor. Datadog charges to ingest and again to index; New Relic and Grafana both fold logs into per-GB data pricing, which is simpler but still scales with volume. A single flat per-GB logs rate with retention included, like Better Stack Logs, is the honest fit when logs specifically are what runs away, priced against your monthly log GB from the estimator.

Watching an AI feature: one bills it separately, two fold it in

The same split shows up the moment you add an AI feature and want to observe it. Datadog gives LLM observability its own meter, a separate line now grouped under Agent Observability, billed per LLM span, where a span is a single call to a model provider. On the annual plan the first 100,000 LLM spans a month run about 160 dollars, each additional 10,000 spans about 3.50 dollars, and it sits on top of the host and log meters rather than inside them. New Relic and Grafana do not break it out at all. New Relic has no separate AI line: the traces your AI calls produce count as ordinary data ingest against the same per-GB rate past the free 100 GB, and access is a normal user seat. Grafana is the same idea on its own axes, where LLM telemetry is just more metric series, log GB and trace GB on the meters you already pay, with no generative-AI line of its own.

So the practical difference is where the AI cost lands, not whether you pay it. On Datadog an AI feature adds a new, visible per-span line you can budget on its own; on New Relic and Grafana it quietly grows a meter you already watch, which is simpler to reason about but easy to miss until ingest or series count climbs. The estimator here sizes the host, APM and log meters, not the per-span AI line, so treat that as a separate number to add on the platform that breaks it out. The sibling guide on why a Datadog bill keeps climbing covers that LLM Observability line in more detail. (List prices, checked July 2026, they move.)

See your own number

Because the three meter on different axes, a like-for-like sticker comparison misleads. The honest way to choose is to put your real host count, full-user count and monthly data volume into the estimator, read which meter dominates, and edit every rate to match a quote before you commit. List pricing is only a baseline; your negotiated or on-demand rate will differ.