JMeter vs LoadRunner comes down to protocols and budget. Apache JMeter is free, open source and covers web, API, database, messaging and mail testing from any operating system with Java. LoadRunner, now sold by OpenText as Professional Performance Engineering, is commercial and supports 180+ protocols and technologies, including SAP GUI, Citrix and mainframe terminals that JMeter does not reach.
For HTTP and API testing, both do the job, and the deciding questions become who writes the scripts, where the load comes from and who pays. The sections below compare them on each of those, using OpenText’s and Apache’s own documentation as of October 2026.
Key Takeaways
- JMeter is free under the Apache License 2.0. LoadRunner Professional and Enterprise are priced through sales; a free Community licence allows 50 virtual users for proofs of concept and non-production testing.
- LoadRunner’s protocol list (SAP, Citrix, Oracle Forms, 3270/5250 terminals, Siebel) is the main reason to pay for it.
- JMeter tests are built in a GUI and saved as .jmx files; LoadRunner scripts are written in VuGen in C, JavaScript or Java.
- LoadRunner’s Controller, VuGen and Analysis install on Windows; JMeter runs wherever Java 8 or later does.
- LoadRunner Professional’s Controller can run .jmx files, but it ignores the thread group settings and supports only Web HTTP samplers.
JMeter vs LoadRunner at a glance
| Apache JMeter | OpenText LoadRunner Professional | |
|---|---|---|
| Current name | Apache JMeter 5.6.3 | OpenText Professional Performance Engineering (version 26.3) |
| Licence | Open source, Apache License 2.0 | Commercial; free Community licence, 50 virtual users |
| Price | Free | Quote from sales (no public list price) |
| Protocols | 10 families on the home page: HTTP(S), SOAP/REST, FTP, JDBC, LDAP, JMS, mail, TCP, native commands, Java objects | ”180+ protocols and technologies” |
| Script creation | GUI test plan (.jmx), HTTP(S) recorder, Groovy via JSR223 | VuGen recording and scripting in C, JavaScript or Java; TruClient for browsers |
| Running tests | CLI mode; distributed client and servers you set up | Controller drives load generators on Windows or Linux |
| Results | HTML dashboard report with APDEX and percentiles | Analysis tool to view, slice and compare runs |
| Operating system | Any OS with Java 8+ (Java 17+ recommended) | Controller, VuGen and Analysis on Windows; load generators on Windows or Linux |
The other two members of the LoadRunner family follow the same renaming: LoadRunner Enterprise is now OpenText Enterprise Performance Engineering, and LoadRunner Cloud is OpenText Core Performance Engineering, a hosted service.
What is the difference between JMeter and LoadRunner?
JMeter is a free Java application maintained by the Apache Software Foundation. You build a test plan from elements in a GUI, run it from the command line, and scale it by adding machines you manage. LoadRunner is a commercial suite of separate tools: VuGen to write scripts, Controller to run scenarios across load generators, and Analysis to study the results, with protocol support far beyond the web.
The practical difference shows up in three places. Protocols: if the system under test speaks SAP GUI, Citrix ICA or a 3270 terminal, LoadRunner supports it and JMeter does not. Cost: JMeter costs only the machines it runs on, while LoadRunner adds licences. Skills: JMeter testers assemble elements in a tree and add Groovy where needed, while LoadRunner scripters write C or JavaScript in VuGen.
How much do JMeter and LoadRunner cost?
JMeter has no licence cost. It is open source under the Apache License 2.0, and every feature is in the download. What you pay for is infrastructure: the load generator machines, their network, and the time to set up and maintain a distributed run.
LoadRunner pricing depends on the product:
- Professional and Enterprise: no public price. The Professional product page offers a free trial with “no credit card required”; both pages route licence questions through “Contact us”. The on-premises licence types include permanent, time-limited and VUFD (Virtual User Flex Day) licences, where one VUFD is a virtual user available for a single 24-hour period.
- Community licence: free and installed with the product, with “50 Vusers for all protocols, including Gatling and JMeter”, except the COM/DCOM, Templates and GUI bundles. OpenText’s bundle sheet limits it to proofs of concept and non-production testing.
- Core Performance Engineering (formerly LoadRunner Cloud): billed by virtual user hours (VUH), where one VUH is one virtual user running for one hour. OpenText’s own AWS Marketplace listing showed $0.15 per Dev VUH, $0.30 per Web VUH and $1.50 per GUI VUH (aws.amazon.com/marketplace, Core Performance Engineering, October 2026), and a 30-day trial includes 1,000 GUI VUH.
A rough way to compare: a Web test with 500 virtual users that runs for one hour in total, ramp-up included, is 500 Web VUH on Core, $150 at the listed rate. The same test on JMeter costs whatever the load generator machines cost you for that hour, plus the time to set them up and run them.
Which protocols does each tool support?
JMeter’s home page lists what it can load test: web (HTTP and HTTPS), SOAP and REST web services, FTP, databases via JDBC, LDAP, message-oriented middleware via JMS, mail (SMTP, POP3 and IMAP, with or without TLS), native commands or shell scripts, TCP and Java objects. Plugins add more. It works at protocol level and, in the project’s own words, “does not execute the Javascript found in HTML pages”.
LoadRunner’s supported protocols matrix for version 26.3 is much longer. Beyond web and APIs it includes SAP GUI, Citrix ICA, Oracle NCA (Oracle Forms), Siebel, RTE (Remote Terminal Emulator) for 3270, 5250 and VT terminals, Windows Sockets, Tuxedo, MQTT and Kafka. It also records browser sessions with TruClient (TruClient 2.0 records in an installed Chrome or Edge), and LoadRunner can run JMeter and Gatling scripts as protocols of their own.
If your test plan is websites and APIs, protocol count is not a reason to choose LoadRunner. If it includes a thick-client ERP, a Citrix-published application or a mainframe green screen, it usually is.
How are JMeter and LoadRunner scripts built?
A JMeter test plan is a tree: a Thread Group for the virtual users, HTTP Request samplers, config elements such as the Cookie Manager, timers for think time, extractors and assertions. You build it in the GUI, often starting from the HTTP(S) Test Script Recorder, and save it as a .jmx file. Custom logic goes in JSR223 elements, and the manual recommends Groovy because it compiles and caches scripts, which BeanShell does not. No programming is needed for a basic test.
LoadRunner scripts are code from the start. VuGen records a business process and generates a script: C for most protocols, JavaScript for Web HTTP/HTML (VuGen uses Chrome’s V8 engine, ECMAScript 2022), or Java for the Java Vuser protocol. OpenText’s product page advertises auto-correlation, where VuGen detects dynamic values in a recording and adds the code to capture them. In JMeter you add extractors yourself, which our guide to correlation in JMeter walks through step by step.
Testers without a programming background usually get productive faster in JMeter. Teams with VuGen experience and a library of existing scripts have a strong reason to stay.
How do you run large tests with each tool?
JMeter runs a load test in CLI mode, and the manual is blunt about it: “Don’t run load test using GUI mode !” To go beyond one machine, a JMeter client controls remote JMeter servers over RMI. The servers must run “exactly the same version of JMeter”, data files such as CSVs “are not sent across by the client”, and each server runs the full plan: 1,000 threads on 6 servers is 6,000 threads.
That setup is work you own: provisioning machines, opening RMI ports, copying data files and keeping versions in step. Our distributed JMeter setup guide covers the steps.
LoadRunner Professional’s Controller “organizes, drives, manages, and monitors load and performance tests”. It sends virtual users to load generators, which install on Windows or on Linux (Oracle Linux, Red Hat, SUSE and Ubuntu are listed for 26.3) and can run in Docker. LoadRunner Enterprise is aimed at distributed testing across several locations, with teams sharing scripts, load generators and results, and Core is the hosted edition, billed by virtual user hour.
How do you read the results?
JMeter writes raw results to a .jtl file and generates an HTML dashboard from it with -e -o after the run, or with -g from an existing file. The dashboard includes an APDEX table, a statistics table with three configurable percentiles, and charts of response times and throughput over time. Each report describes one run, so comparing runs is up to you.
LoadRunner’s Analysis is a separate application built for exactly that: it lets you “view, dissect, and compare the results of the tests”. Having that comparison built in, rather than assembled from separate reports, is one of the things LoadRunner licences pay for.
Which operating systems do JMeter and LoadRunner run on?
JMeter is a “100% pure Java application” and runs on any operating system with Java 8 or later; the 5.6 changes page recommends Java 17. You can write a test on a Mac and run it on Linux load generators without changes.
LoadRunner’s 26.3 system requirements list the full installation (including Controller), VuGen and Analysis under Windows only: Windows 10 and 11 and Windows Server 2016 to 2025. Linux appears for the standalone load generator and two helper components. The DevWeb protocol is the cross-platform exception, described as running on Windows, Linux and macOS.
Can LoadRunner run JMeter scripts?
Yes, with conditions. Professional’s Controller runs .jmx scripts side by side with other protocol scripts, on Windows or Linux load generators, using a built-in JMeter 5.6. But it ignores the thread group’s Number of Threads, Ramp-up period and Loop Count, and you configure those in Controller instead. A script with several thread groups must be split into one script per group, only Web HTTP samplers are supported, and the tearDown Thread Group is not. LoadRunner Enterprise runs only the first thread group of a multi-group script, and Core creates a copy of the script for each enabled thread group and supports JMeter 3.0 to 5.6.
That makes LoadRunner a reasonable place to run simple HTTP .jmx files alongside VuGen scripts, and a poor fit for JMeter plans that depend on non-HTTP samplers or carefully tuned thread groups.
When to choose JMeter and when to choose LoadRunner
Choose JMeter for web, API and JMS testing on a zero licence budget; choose LoadRunner when the system speaks SAP GUI, Citrix or a terminal protocol. In more detail, choose JMeter when:
- You test websites, REST or SOAP APIs, databases over JDBC, or JMS queues.
- You want no licence cost and plain .jmx test files you can keep in version control.
- Your team works on macOS or Linux, or runs tests in CI containers.
- You prefer building tests in a GUI over writing code.
Choose LoadRunner when:
- You must test SAP GUI, Citrix, Oracle Forms, Siebel or terminal-based mainframe applications.
- You need a commercial vendor behind the tool.
- Your team has years of VuGen scripts and the skills to maintain them.
- You want correlation and analysis tooling built in rather than assembled.
Plenty of organisations run both: LoadRunner for the ERP and the green screens, JMeter for the web and API layer. If you are deciding between JMeter and the newer code-first tools as well, our k6 vs JMeter vs Gatling vs Locust comparison covers that side.
Running JMeter at scale without managing load generators
The part of JMeter that costs real effort is distributed execution: machines, ports, data files and matching versions, for every run.
JMeter cloud load testing on LoadFocus runs your existing .jmx unchanged on managed load generators in 25 cloud regions: upload the plan and its CSV files, then set virtual users, duration and ramp-up in the browser. Thread groups, timers, assertions and JSR223 samplers run as they do on your machine, the Enterprise plan goes up to 20,000 virtual users for up to 5 hours, and results break response times and errors down by region.
If you are moving off LoadRunner specifically, the LoadRunner alternative page compares the two products directly.
Frequently Asked Questions
Is JMeter better than LoadRunner?
For web and API load testing, JMeter does the same job without a licence fee and runs on any operating system with Java. LoadRunner is better when you need protocols JMeter lacks, such as SAP GUI, Citrix or 3270 terminals, or vendor support and its Analysis tool. Neither is better across the board.
Is LoadRunner free?
LoadRunner has a free Community licence with 50 virtual users for all protocols except the COM/DCOM, Templates and GUI bundles, meant for proofs of concept and non-production testing. Full LoadRunner Professional and Enterprise licences are priced through OpenText sales, and the hosted Core edition is billed per virtual user hour.
What is LoadRunner called now?
OpenText renamed the family: LoadRunner Professional is now OpenText Professional Performance Engineering, LoadRunner Enterprise is OpenText Enterprise Performance Engineering, and LoadRunner Cloud is OpenText Core Performance Engineering. The documentation for version 26.3 uses the new names.
Can LoadRunner run JMeter scripts?
Yes. LoadRunner Professional’s Controller can run .jmx files with a built-in JMeter 5.6, but it ignores the thread group’s threads, ramp-up and loop count, supports only Web HTTP samplers, and asks you to split a plan with several thread groups into separate scripts.
Does JMeter require coding?
No, not for most tests. You build a JMeter test plan in the GUI from thread groups, samplers, timers and assertions, or record it with the HTTP(S) Test Script Recorder. Code is optional, through JSR223 elements written in Groovy, for custom logic such as signing requests or generating data.
Which is easier to learn, JMeter or LoadRunner?
JMeter is usually easier to start with: it is a free download, and a first test needs no programming. LoadRunner asks you to learn VuGen scripting in C or JavaScript, plus Controller and Analysis, though its recording and auto-correlation reduce the manual work once you know the tools.
Sources
- Apache JMeter home page: protocols, open source, “does not execute the Javascript”, checked 2026-10-25.
- JMeter download page: version 5.6.3, requires Java 8+ and changes: Java 17 or later recommended, checked 2026-10-25.
- JMeter user manual, Getting Started: CLI mode and Best Practices: JSR223 and Groovy, checked 2026-10-25.
- JMeter user manual, Remote Testing and Generating Report Dashboard, checked 2026-10-25.
- OpenText Professional Performance Engineering product page: “180+ protocols and technologies”, the new product names, free trial: opentext.com/products/professional-performance-engineering, October 2026.
- LoadRunner Professional 26.3 help: Welcome (VuGen, Controller, Analysis), License Utility (Community licence, VUFD), JMeter tests in Controller: admhelp.microfocus.com, October 2026.
- Supported Protocols 26.3 and System Requirements 26.3: admhelp.microfocus.com/documents/lr, October 2026.
- VuGen 26.1-26.3 help: Vuser types, JavaScript Web HTTP scripts: admhelp.microfocus.com/vugen, October 2026.
- LoadRunner Enterprise 26.1-26.3 help (JMeter scripts) and Core Performance Engineering 26.3 help (VUH licences, trial, JMeter scripts, support matrix): admhelp.microfocus.com, October 2026.
- Core Performance Engineering (LoadRunner Cloud), sold by OpenText: AWS Marketplace listing, October 2026.
- OpenText Professional and Enterprise Performance Engineering licence bundles (PDF, 08.26): opentext.com, October 2026.
