Amazon CloudWatch pricing guide details
Pricing last verified: April 2026. Plans and pricing may change — check the vendor site for current details.
Pricing Overview
Amazon CloudWatch uses a pay-as-you-go freemium model. There's a free tier that covers small workloads for $0, then incremental usage is billed per-metric, per-GB of log ingestion, per-alarm, and per-query. AWS publishes per-unit rates, not monthly plans: there is no lowest monthly price and no top one, so the same product costs a hobby developer nothing and a large enterprise thousands of dollars for fundamentally the same service.
This pricing model is typical of AWS products: no upfront commitment, no minimum spend, no seat-based licensing. It rewards disciplined usage (scoped log retention, deliberate alarm counts, metric granularity only where you need it) and punishes "turn everything on" defaults — particularly around log ingestion and Logs Insights query volume, which are the two surprise cost drivers most teams hit first.
Plan Comparison
CloudWatch doesn't have named tiers like "Basic / Pro / Enterprise" — it's consumption-based, and AWS publishes a rate per unit of telemetry, all billed on the same AWS invoice:
| What is charged | Published rate | Notes |
|---|---|---|
| Free tier | Included with AWS account | 10 custom or detailed-monitoring metrics, 10 alarm metrics, 1M API requests, 5GB of Logs data covering ingestion, archive storage and Logs Insights scans, 3 dashboards, 1,800 Live Tail minutes |
| OpenTelemetry metrics | $0.50 per GB ingested | Flat rate including 15 months of storage; no separate charge per API call or metric series |
| PromQL queries | $0.01 per million samples scanned | Console and dashboard queries are free |
| PutMetricData | $0.01 per million requests | On top of the metric storage charge |
| Classic custom metrics | $0.30 / $0.10 / $0.05 per metric | Tiered at the first 10,000, the next 240,000, and beyond 250,000. AWS states these only inside its worked examples |
| Live Tail | $0.01 per minute | After the free 1,800 minutes a month |
| Metric centralization | $0.05 per GB copied | First copy to a destination account is free |
| Cross-account observability | No additional charge | Query across accounts without copying data |
Rates are US East (N. Virginia) and vary by region.
Unlike seat-based tools, CloudWatch bills only on what you use — three developers and three hundred developers pay the same if their workload generates the same telemetry. This makes it one of the cheapest observability tools for sparse workloads and one of the most expensive for verbose logging without retention discipline.
Hidden Costs and Considerations
Cross-account observability lets you view and query metrics and logs across accounts from a central monitoring account without copying data. There is no additional charge.
Cost Estimates by Team Size
These are rough monthly estimates for AWS-centric workloads, assuming moderate discipline on log retention:
- Solo developer / hobby project: $0 – $5. Free tier usually covers everything; cost appears only with custom metrics or detailed monitoring on several resources.
- Small team (5 engineers, 1 production environment): $30 – $150. Typically driven by 20-50 GB of log ingestion plus custom metrics and a handful of Container Insights-enabled services.
- Mid-size team (20 engineers, multiple environments): $300 – $1,500. Cross-account setup plus multiple EKS clusters with Container Insights plus higher log volume. The Logs Insights query bill starts to matter at this scale.
- Large enterprise (100+ engineers, many accounts): $3,000+, with no ceiling. Spend is driven by log ingestion volume, retention, and query load across accounts rather than by any published maximum; teams typically pair CloudWatch for collection with Datadog or Splunk for analysis.
A practical rule: start with the free tier, add Container Insights selectively, and set log retention explicitly on every log group. Most teams overspend by 2-3× before they tighten retention policies.
How Amazon CloudWatch Pricing Compares
CloudWatch's pricing structure is unusual in observability because it has no per-host or per-user minimum:
- Datadog: Free tier plus paid plans starting at $0.75 per host per month (plus usage-based add-ons). Per-host billing means even a small team with many hosts pays meaningfully more than CloudWatch at low telemetry volumes, but Datadog absorbs multi-cloud and application-layer workloads CloudWatch can't.
- New Relic: Free tier plus paid plans starting at $19 per month per host. Most expensive entry point among the three, but the per-user pricing for platform access can work out cheaper than Datadog for small teams with many hosts.
- Grafana Cloud: Freemium with enterprise pricing on request. Grafana Cloud plus CloudWatch via Metric Streams is a common mid-ground — CloudWatch collects, Grafana visualizes, you pay only for storage and queries.
The honest summary: if your workload lives on AWS, CloudWatch is almost always the cheapest floor. Competitors make sense when you outgrow its dashboards, need multi-cloud, or want application-layer observability as a primary workflow.