What is API load testing?

API load testing measures how a service behaves when many clients call it at the same time. It works at the protocol level, sending HTTP requests directly rather than driving a browser, so it answers questions about throughput, response time and error rate under concurrency. It is the standard way to size an API before a launch, a sale, or a migration.

What is the difference between load, stress, spike and soak testing?

Load testing holds an expected level of concurrency and checks that response times stay acceptable. Stress testing pushes past the expected level to find where the service breaks. Spike testing applies a sudden jump in traffic to see how the system absorbs it. Soak testing holds a moderate load for hours to surface memory leaks and resource exhaustion. All four are shaped by the same controls: user count, ramp-up and duration.

Which tool do I write my API load test in?

Whichever your team already uses. TestMu AI does not replace JMeter or Gatling, it runs them. Apache JMeter test plans in .jmx form and Gatling Java DSL simulations are both supported natively and run without a rewrite. There is no proprietary scripting language to learn and no plan to re-validate.

Which protocols can I test?

Whatever your existing plan already exercises. Protocol coverage is a property of JMeter and Gatling rather than of the runner, so a plan that drives REST endpoints today keeps driving them unchanged. TestMu AI provides the load generators, the regional distribution and the reporting around that plan.

Can I run API load tests from my CI pipeline?

With Gatling, yes. Gatling runs from a hyperexecute.yaml committed alongside the simulation in your own repository, so the run configuration is versioned with the test. JMeter is configured through the Projects portal and has no CLI or YAML path, so a pipeline-triggered load stage should use Gatling.

How is test data spread across load generators?

Split CSV divides the input file so each generator reads a unique slice. Without it every machine starts at row one and replays the same identities, which exercises your caches instead of your service. On a run with nine generators and a 10,000 row file, the split gives each machine its own range and no row is reused.

How many virtual users can I generate?

Concurrency is a plan entitlement: 100 concurrent virtual users on Free, up to 2,000 on Basic, up to 8,000 on Pro, and a custom figure on Enterprise. What each virtual user does decides what you reach inside that ceiling. Browser-driving samplers lower it, as do missing HTTP timeouts, large payloads, and long chained workflows. Lightweight HTTP samplers with explicit timeouts reach considerably more than heavy chained journeys.

Where does the load come from?

Six named regions are documented: East US in Richmond, Virginia, which is the default, West US 2 in Moses Lake, Washington, Central India in Pune, Southeast Asia in Singapore, Brazil South in Sao Paulo State, and Mexico Central in Queretaro State. Traffic is split as a percentage of total users, so you can weight a run toward the geography your customers are in. Multi region load generation is an Enterprise plan entitlement; Free, Basic and Pro run single region.

Does this replace functional API testing?

No, and the two answer different questions. Load testing tells you whether the service stays fast and available under concurrency. Functional testing tells you whether the response is correct. Run functional API checks on the automation cloud and load runs on HyperExecute, against the same build.

What do I get back after a run?

Five reports: a Summary Report with virtual users, ramp-up, average response time, throughput, bandwidth, regions and duration, a Timeline Report, Request Stats broken down per request, an Errors report grouping codes and counts with their sample distribution, and Logs. A troubleshooting table maps common failure modes to the report that explains them.

API Load Testing

Protocol level load runs driven by the JMeter or Gatling plan you already maintain. TestMu AI provisions the generators, drives the run, and tears it all down.

Start free with Google

Start free with Email

Automate Browser Flows from your
Terminal with Kane CLI

Explore Kane CLI

Trusted by 3M+ users globally at

+Read case study

Four Shapes of the Same Run

Load, stress, spike and soak are not different products. They are the same plan shaped by user count, ramp-up and duration, which are three of the eight fields on the run form.

Load

Hold the concurrency you expect on your busiest day and confirm response times stay inside budget.

Steady users, fixed duration

Stress

Push past the expected level until something gives, so you learn the breaking point before your customers do.

Rising users, no ceiling

Spike

Jump straight to peak with almost no ramp-up, the way a campaign launch or a sale actually arrives.

Short ramp-up, high users

Soak

Hold a moderate level for hours to surface memory leaks, connection pool exhaustion and slow degradation.

Moderate users, long duration

The Execution Layer for the Plan You Already Have

TestMu AI does not replace JMeter or Gatling. It runs them, on infrastructure that exists only while the test does.

Start free with Google

MANAGED RUN

Provision, Run, Report, Destroy

Load generators start cold at the beginning of a run and are destroyed at the end. Nothing idles between tests.

Start free

No load generation fleet to patch or keep alive.
Nine machines across three regions on a single run.
Reports and logs collected against the job automatically.

SPLIT CSV

Give Each Generator Real Data

Split CSV hands every generator a unique slice of your input file, so no two machines replay the same rows.

Start free

Unique identities per generator, not duplicated sessions.
Ten thousand rows divided across nine machines.
Without the split you measure cache, not your service.

REQUEST STATS

Read Results Per Endpoint

Request Stats breaks the run down per endpoint, and the Errors report groups failures by response code.

See the reports

Samples, average response time and throughput per route.
Errors grouped by code with their sample distribution.
Summary, Timeline, Request Stats, Errors and Logs per run.

Pick the Tool. We Run It.

Start free with Google

The plan formats that run on TestMu AI today, and where each one fits.

Plan format

Runs natively on TestMu AI

Where it fits

Apache JMeter

.jmx

Yes, portal project type

Large existing plans, GUI authoring, CSV data

Gatling

Java DSL .java

Yes, portal and hyperexecute.yaml

Load as code, versioned with the service

JavaScript

Documented, YAML and CLI driven

Script-first teams already on k6

Load Testing Inside One Quality Platform

The grid running your API load plan is the grid running your functional suites, under one account and one bill.

HyperExecute

Automation Cloud

KaneAI

Test Intelligence

HyperExecute

The orchestration cloud that provisions load generators, drives the run, and destroys the infrastructure after.

More about HyperExecute

Two Plan Formats

JMeter .jmx and Gatling Java DSL simulations run natively.

Six Load Regions

Split traffic by percentage across named physical locations.

No Idle Fleet

Generators exist only for the duration of the run.

Performance Testing Plans and Limits

Start free with Google

Concurrency, virtual user hours, load generators and test duration are all plan entitlements. Verified on the pricing page on 8 September 2026.

Free

Basic

Pro

Enterprise

Price, annual

$79 / month

$349 / month

Custom

Price, monthly

$99 / month

$449 / month

Custom

Concurrent virtual users

100

Up to 2,000

Up to 8,000

Customizable

Virtual user hours / month

100, introductory

2,000, introductory

8,000

Customizable

Load generators

Up to 2

Up to 20

Customizable

Max test duration

40 minutes

2 hours

6 hours

Customizable

Load regions

Single region

Multi region

Some Love from our Customers

As Best Egg expanded its product offerings and entered new markets, we knew our old testing infrastructure couldn’t keep up.
With support from Tenny Agustin, our Engineering Operations Lead, we modernized our approach with

TestMu AI

Best Egg

best-egg

Excited to Share My Learning Journey with Kane AI & Lambda Tool!
I'm pleased to announce that I've recently gained hands-on experience exploring Kane AI through the Lambda Tool and it’s been a fantastic journey of upskilling!

KaneAI

Suryateja Goud

suryateja-goud

See how is #Futureready to enable blazing-fast test orchestration seamlessly integrated with organizations' existing CI/CD platforms, using #Microsoft Azure.

TestMu AI