Datadog's Free Plan and Trial - The 5-Host Limit and Where Paid Begins

This article sorts out how far you can use Datadog for free, separating the Free plan from the 14-day trial. It confirms what is and is not included in Free using the comparison table on the pricing page, and covers the "moments when free breaks" such as the sixth host and 1-day retention, the first SKU to choose when going paid, and the verification steps to complete during the free period.

Conclusion first: the Free plan and the 14-day trial are different things

The question "can Datadog be used for free?" has two separate answers. The first is the Free plan, a permanent free tier that the pricing page describes, as of September 2026, as "free to start, core collection and visualization features, 1-day metric retention, up to 5 hosts" (source: Datadog Pricing). The second is the free trial, which the official trial page describes as "try Datadog for 14 days," "no credit card required," and "explore the full platform for 14 days" (as of September 2026; source: Free Datadog Trial).

Differences between the Free plan and the trial (as described on the official pages as of September 2026)
ItemFree plan14-day trial
DurationNo expiry14 days
CostFreeFree (no credit card required)
Number of hostsUp to 5No limit stated (described as "unlimited monitoring")
Metric retention1 dayNo statement explicitly confirming the Pro-equivalent 15 months could be found
Usable scopeCore collection and visualizationThe full platform
Sources: Datadog Pricing / Free Datadog Trial, cited above

Confusing these two produces the misunderstanding that "alerts and APM worked during the trial, but once dropped to Free nothing works." If you accept that Free is a tier for "looking at 5 hosts for one day's worth of data" and the trial is a tier for "seriously testing part of production for just two weeks," the way to use each becomes clear. This article confirms the scope and boundaries of these two using only the pricing page comparison table and the statements in the official documentation. This article explains how billing works. For current unit prices and plans, refer to Datadog's official pricing page.

The scope of the Free plan - what the comparison table shows you can and cannot use

The Infrastructure Monitoring comparison table on the pricing page lists feature availability in three columns: Free / Pro / Enterprise. Reading the Free column as of September 2026, standard events and metrics for up to 5 hosts, standard dashboards, more than 1,000 Integrations, host map and container map, and unlimited users are checked, and full-resolution data retention is 1 day. On the other hand, customizable alerts, the Container Monitoring allotment, the custom metrics allotment, custom events, single sign-on via SAML, and outlier detection have no entry in the Free column (source: Datadog Pricing).

The scope of the Free plan as read from the Infrastructure Monitoring comparison table (as of September 2026)
Included in FreeNo entry in the Free column (Pro or higher)
Standard events and metrics (up to 5 hosts)Customizable alerts (unlimited on Pro / Enterprise)
Standard dashboardsContainer Monitoring allotment (Pro 5 / Enterprise 10 containers per host)
More than 1,000 integrationsCustom metrics (Pro 100 / Enterprise 200 per host)
Host map and container mapCustom events (Pro 500 / Enterprise 1,000 per host)
Unlimited usersSAML single sign-on, outlier detection
1-day data retention15-month data retention
Source: the Infrastructure Monitoring comparison table in Datadog Pricing

What matters most in practice is the first row of the right column. Because Free has no check in the customizable alerts row, you cannot assume the basic monitoring action of "notify Slack when CPU exceeds 90%." It is more accurate to see the Free plan not as a "monitoring tool" but as an "observation window for confirming what becomes visible once you install the Agent," and for that purpose 5 hosts / 1 day is plenty. The experience of standard dashboards filling up the moment you install the Datadog Agent, the feel of configuring integrations, and the handling of narrowing down by tags can all be confirmed within this tier.

What the trial unlocks, and what could be confirmed about what happens after it ends

The trial page makes three statements, "the full platform for 14 days," "no credit card required," and "14 days of unlimited monitoring," and contains no wording that limits the products covered (as of September 2026; source: Free Datadog Trial). In addition, for organizations already under contract there are trials of individual products, and the official documentation explains that the start and end dates can be checked on the Trial Management tab of the Plan & Usage screen, that there is no charge during the trial, and that usage is shown on the Usage & Cost tab as "estimated cost had there been no trial" (source: Plan & Usage). This "display of estimated cost" is the most valuable feature of a trial: run production-equivalent volume for two weeks and a projection of your bill appears on screen before you sign a contract. On the other hand, to be candid, no official page explicitly stating how the account is handled when the initial 14-day trial ends (whether it automatically moves to the Free plan or you are asked to choose a paid plan) could be found in the research at the time of writing. Two facts could be confirmed. Since no credit card registration is required, no means is on file for billing to start automatically the moment the trial ends. And the official documentation explains to administrators that "you upgrade or downgrade your account from the Plan tab of Plan & Usage" (same source). Therefore, confirm the behavior at the end of the trial with your own eyes, through the terms displayed on screen when you start the trial and what the Plan tab shows. This article does not fill that gap with guesswork.

The scale at which "staying free" works, and the moments it breaks

Operating on the Free plan is viable when three conditions coincide: 5 or fewer monitored hosts, you only want to see the "current" state, and you do not need notifications. A personal test environment, or observing a few servers hosting internal tools, fits these conditions. There are three moments when it breaks. First, the sixth host. The official billing documentation defines a host as "a physical or virtual OS instance monitored by Datadog" and states explicitly that Kubernetes nodes and Azure App Service instances each count as one (as of September 2026; source: Billing). Just by spinning up one development EC2 instance, or because EC2 instances detected by the AWS integration are counted, the 5-host limit is easily exceeded. The saving grace is the statement in the same document that "hosts that stop reporting (INACTIVE in the infrastructure list) are not counted for billing and take up to 2 hours to disappear from the list," so a host you have stopped is not counted indefinitely. Second, 1-day retention. Retention of 1 day means that the basic monitoring actions of "compare with the same time slot last week" and "see whether it has kept growing since last month" are not possible. The day after an incident, yesterday's evidence is gone. Third, notifications. As noted in the previous section, the Free column has no check for customizable alerts, so the only ones who notice an anomaly are the people looking at the screen.

  • The sixth host - a test EC2 instance or AWS integration detection pushes you over the limit. Once INACTIVE, it drops out of the count within 2 hours
  • 1-day retention - no day-over-day / week-over-week comparison, and the evidence is gone the day after an incident
  • No alerts - the only means of noticing is human eyes, leaving nights and holidays undefended
The three moments when Free plan operation breaks (basis = the pricing page comparison table and billing documentation as of September 2026)

The point at which any of these three turns into "a problem" is the signal to consider going paid. Conversely, in an environment where none of the three is a problem, there is no reason to force a move to paid.

Where paid begins - which SKU and which contract form to start with

The first step into paid is, almost without exception, Infrastructure Monitoring Pro. The official pricing page lists it as a monthly amount per host with annual-contract and on-demand unit prices side by side, and this single SKU extends retention to 15 months, makes alerts unlimited, and includes 100 custom metrics and Container Monitoring for 5 containers per host (source: Datadog Pricing). Enterprise is one step up in host unit price and adds Watchdog automatic detection, anomaly detection, Live Processes, and so on, but there is usually no reason to choose it for a first contract. As for the contract form, the official documentation explains two billing plans. One is the High Water Mark Plan (HWMP), which measures the host count hourly and bills at month end on the 99th percentile maximum. It states explicitly that the top 1% is excluded to limit the effect of short spikes. The other is the Monthly / Hourly Hybrid Plan (MHP), in which you set a monthly minimum commitment and host hours beyond it are billed at an hourly rate; it is described as suited to transient environments such as autoscaling (as of September 2026; source: Pricing (Billing)). The billing cycle begins on the 1st of every month (UTC) regardless of the signup date, with the first month prorated according to the signup date (source: Billing, cited above). The practical line at the boundary is as follows.

The first choice when moving from Free to paid
SituationFirst choiceShape of the monthly cost
6 to 10 hosts and you want notificationsInfrastructure Monitoring Pro on-demandOn-demand unit price × number of hosts
Host count is predictable and you will use it for six months or moreInfrastructure Monitoring Pro on an annual contractAnnual-contract unit price × number of hosts
Host count fluctuates dailySet a monthly minimum commitment on MHP and put the fluctuating portion on hourly billingUnit price for the committed portion + hourly rate for the excess (rates depend on the contract)
Source for billing units: Datadog Pricing. For unit prices, refer to the same page. Other products such as logs / APM are not included

The gap between the annual and on-demand unit prices looks small per host, but it accumulates in proportion to the host count and the number of months. Even so, to keep to the order of not signing an annual contract before measuring, it is safest to pay on-demand for the first one or two months. How to stack up other products is covered in Decoding the Pricing Structure. This article explains how billing works. For current unit prices and plans, refer to Datadog's official pricing page.

Three jobs to complete during the free period, and the bridge to the adoption decision

The Free plan and the trial are tiers for "trying," but if you finish with just trying, rework happens right after going paid. There are three jobs which, if completed during the free period, make billing and operations stable from the start after signing.

  • 1. Decide the tag convention and bake it into the Agent - fix the names and notation of env / service / team while you are still at 5 hosts, put them into the Agent configuration, and only then add hosts
  • 2. Learn how to watch usage - build the habit of looking at hourly usage in the Usage reports under Plan & Usage, and at the top 5,000 custom metrics by submission count
  • 3. Measure your submission volume - run production-equivalent logs and custom metrics during the trial and use the estimated cost under Usage & Cost as your pre-contract estimate
The three jobs to complete during the free period (basis = the Plan & Usage documentation as of September 2026)

The basis for 2 and 3 is the Plan & Usage explanation in the official documentation: administrators can check hourly usage for every billable parameter and the top 5,000 custom metrics submitted, and usage of trial products is shown as estimated cost without being billed (source: Plan & Usage). Since paper calculations of monitoring tool costs almost never hit the mark, using these "two weeks of free measurement" as your estimation period is the most reliable way to deal with Datadog's pricing structure. The tag convention in 1 is the foundation that affects everything, from filtering to dashboards to the scoping of Monitors, and fixing it later means rewriting all of them. Once the three are done, the next step is the decision of "whether to adopt." How to think by scale band and the six items to decide before signing are in the adoption checklist, and how to check usage on an ongoing basis is summarized in the Metrics explanation. Use the free tier for "measurement," not just for "experience." That alone greatly lowers the odds of being surprised by your first paid invoice.

References

Share