Application performance monitoring tools
Application performance monitoring (APM) tools trace requests through your code and services so you can find slow endpoints, errors, and bottlenecks before customers do. This grid compares ten widely used platforms, from all-in-one suites to open-source-friendly and developer-first tools, on what they cost to start, how they charge, free plans, and the capabilities that most often change a choice.
Last updated
How to choose
Start with how you will be billed, because that is where APM bills surprise people. Per-host plans (Datadog, Dynatrace, Splunk, AppDynamics) are predictable for stable fleets but expensive for many small containers. Per-gigabyte plans (New Relic, Elastic, SigNoz, Grafana) track how much telemetry you send, so sampling and log volume matter more than headcount.
Questions to ask
- How much data will we send? Estimate spans, logs, and metrics per month before comparing, and check what is included free.
- Who needs a login? Some vendors charge per user; others include unlimited seats.
- Do we want to self-host? Open-source-friendly tools can run on your own infrastructure, which changes the cost and the work.
- Are we standardizing on OpenTelemetry? Check native OTLP ingest so you are not locked into a vendor agent.
- What else do we need beside tracing? Logs, real user monitoring, synthetics, profiling, SLOs, and AI features are often priced separately.
- How long is data kept? Default trace retention ranges from a week to a month; longer retention usually costs more.
Pitfalls
List prices rarely include everything you will use. Add the cost of logs, indexed spans, extra retention, and add-ons before comparing, and run a trial with real traffic. Published prices change often; confirm on the vendor's page before you commit.
Answer
Browse the table. Filter to what matters.
Use the filters below, then sign in if you want to shortlist and decide with friends.
Sign in to vote, hide services and features, and reorder.
Verified: hover a cell and click its green dot for the source Estimate: no cited source Not stated: the company doesn't publish it
| Feature | ⠿ | ⠿ | ⠿ | ⠿ | ⠿ | ⠿ | ⠿ | ⠿ | ⠿ | ⠿ |
|---|---|---|---|---|---|---|---|---|---|---|
|
⠿
Lowest monthly cost to start
|
Doesn't apply
|
|||||||||
|
⠿
Entry paid pricing
|
||||||||||
|
⠿
Pricing model
|
||||||||||
|
⠿
Free plan
|
|
|
|
|
|
|||||
|
⠿
Free plan limits
|
Doesn't apply
|
Doesn't apply
|
Doesn't apply
|
Doesn't apply
|
Doesn't apply
|
|||||
|
⠿
Per user fees
|
|
|
|
|
|
|||||
|
⠿
Self hosting option
|
|
|
|
|
|
|
|
|
||
|
⠿
OpenTelemetry ingest
|
|
|
|
|
|
|||||
|
⠿
Continuous profiling
|
|
|
|
|
||||||
|
⠿
Real user monitoring
|
|
|
||||||||
|
⠿
Synthetic monitoring
|
|
|
|
|
||||||
|
⠿
SLOs
|
|
|
|
|
|
|
|
|||
|
⠿
AI and LLM observability
|
|
|
||||||||
|
⠿
AI debugging assistant
|
|
|
|
|||||||
|
⠿
Trace data retention
|
60 days
|
8 days full-fidelity traces
|
Why these features matter
-
Lowest monthly cost to start ($/month)
Shows the smallest commitment needed to try a vendor with real tracing. Usage and add-ons come on top, so use it to compare entry points, not final bills.
-
Entry paid pricing
The unit you are billed on, more than the sticker price, decides how your bill grows with traffic, hosts, or data.
-
Pricing model
Per-host pricing suits steady fleets; per-GB pricing suits teams that control data volume with sampling; per-user pricing penalizes large teams.
-
Free plan
A free plan lets a small team or side project get real tracing without a card, and lets you test with production-like data.
-
Free plan limits
Free plans differ widely in how far they go before you must pay.
-
Per user fees
Per-seat fees raise the cost of giving everyone on-call access and discourage sharing dashboards.
-
Self hosting option
Self-hosting helps with data residency, cost control at scale, and avoiding lock-in, but adds operations work.
-
OpenTelemetry ingest
OpenTelemetry keeps instrumentation portable so you can change vendors without re-instrumenting.
-
Continuous profiling
Profiling finds hot code paths that traces alone cannot, which cuts compute cost and latency.
-
Real user monitoring
Connects what customers actually experience to backend performance.
-
Synthetic monitoring
Synthetics catch outages and regressions before real users hit them.
-
SLOs
SLOs turn raw metrics into reliability targets that guide when to ship and when to fix.
-
AI and LLM observability
Matters if you ship AI features and need to debug prompts, latency, and spend.
-
AI debugging assistant
Can shorten incident response, though quality and pricing vary.
-
Trace data retention
Short retention limits investigating slow regressions and older incidents.
About the services
-
Datadog APM · Datadog
Full observability suite with APM sold per host plus per-GB span ingestion and indexed spans. Strong breadth (profiling, RUM, synthetics, LLM tracing), but bills grow quickly with hosts and indexed volume.
-
New Relic · New Relic
All-in-one platform priced by data ingested and by user type rather than by host. A generous free tier (100 GB a month) suits small teams, but full-platform users are charged per seat.
-
Dynatrace · Dynatrace
Enterprise full-stack observability with automatic instrumentation and AI-driven root cause analysis. Priced per host and per GiB of telemetry under an annual commitment, with no per-seat fees.
-
Grafana Cloud · Grafana Labs
Hosted Grafana stack (Tempo, Loki, Mimir, Pyroscope) with an always-free tier and usage-based Pro plan. Good fit for teams already on Grafana, Prometheus, or OpenTelemetry who also want the option to self-host.
-
Elastic Observability · Elastic
Observability built on Elasticsearch, offered as serverless, hosted, or self-managed. Serverless pricing is per GB ingested and retained, which rewards teams that already search and analyze data in Elastic.
-
Sentry · Sentry
Developer-first error monitoring with tracing and profiling added. A free single-user plan and a low-cost Team plan make it an easy first APM for small product teams; less suited to infrastructure monitoring.
-
Honeycomb · Honeycomb
Event-based observability built for debugging complex distributed systems, with unlimited users and queries. Priced by event volume, so cost is predictable, but it has no host or infrastructure agent.
-
SigNoz Cloud · SigNoz
OpenTelemetry-native, open-source observability with a hosted cloud and a free self-hosted edition. Pricing is a flat per-GB rate with no user or host charges, aimed at teams leaving Datadog-style bills.
-
Observability Cloud · Splunk
Splunk's SaaS observability suite with host-based pricing, APM with always-on profiling, and full-fidelity tracing. The APM edition starts at $60 per host a month; real user monitoring needs the top tier.
-
AppDynamics · Splunk
Business-focused APM for traditional, three-tier, and on-premises applications, now part of Splunk and Cisco. Editions are priced per host with add-ons for real user monitoring and synthetics.
Suggest a change
Related comparisons
-
Translation management softwareTranslation management systems (TMS) keep a team's UI strings, documents and translations in one place, with translation memory, glossaries, machine and human translation, and integrations with code and design tools. The decision is how you are charged (words, strings or keys), what a free plan covers, and how much automation and security you need.
-
Feature flag toolsFeature flag tools let engineers switch features on or off, roll them out to a slice of users, and roll them back without redeploying. This grid compares ten hosted and self-hosted tools, from open source servers to full experimentation platforms. The real trade-offs are how each one bills (per seat, per request, per user), whether you can run it yourself, and whether approvals, SSO, and A/B testing sit behind an enterprise plan.
-
Incident management softwareIncident management software wakes the right person when something breaks, gives the team one place to run the response, and records what happened so it doesn't happen again. This grid compares eight tools on what most buyers decide on: what the first paid plan really costs, whether on-call paging is included or sold separately, how well incidents run inside Slack or Teams, and how far AI goes.
-
Status page softwareStatus page software gives your customers one trusted place to see whether your service is up, read incident updates, and subscribe to alerts. Some tools are stand-alone status pages, while others bundle uptime monitoring, on-call paging, or full incident response. The trade-offs are price model (flat plan, per user, or per component), how many subscribers and team members each plan allows, and whether monitoring and paging come with it.
-
API management platformsAPI management platforms sit in front of your APIs to handle authentication, rate limits, analytics, a developer portal, and increasingly AI and MCP traffic. This grid compares seven well-known options on what you can see before talking to sales: entry price, free access, how pricing works, where it can run, and which capabilities come built in.
-
CI/CD platformsCI/CD platforms build, test, and ship your code automatically every time you push a change. The real decision is whether to use the pipeline built into the place where your code already lives, or an independent service, and how you want to pay: per user, per build minute, or with your own servers.