QAM Hub vs Zephyr (formerly Zephyr Scale): Test Management Compared

QAM Hub bills for the people who test and keeps automation results, flaky-test history and coverage in one standalone TMS. Zephyr bills on your whole Jira user count and caps its no-code automation runs per month.

Facts verified on 1 October 2026

QAM Hub is billed per tester, with one free read-only viewer for every paid seat, and it stores the results of the Playwright, Cypress or JUnit XML tests you already run next to the test cases they cover. Zephyr is a Jira app, so its price follows your whole Jira user count, and its built-in no-code automation comes with a monthly execution allowance set by your Jira tier. If most of your Jira users test and you want record-and-playback automation inside Jira, Zephyr is cheap and convenient. If your Jira instance is several times larger than your QA team, or your automation already lives in a code framework, a standalone test management system costs less and keeps run history without a monthly allowance deciding how often you can test.

Which Zephyr this page compares#

Half the confusion in Zephyr comparisons comes from names, so this comes first. SmartBear consolidated its Jira test management apps into one product family called Zephyr, and the vendor's own documentation now says it plainly: "We've recently consolidated Zephyr Squad and Zephyr Scale into a single product family: Zephyr." No date for the change is published.

This page compares QAM Hub with the first row, Zephyr Standard and Zephyr Advanced, because that is what a team searching for "Zephyr Scale" is actually evaluating today. Essential comes up where the difference matters, mostly around AI. Zephyr Enterprise is a different product with a different buyer, and we do not cover it here.

One more ownership fact worth having before you shortlist: SmartBear also owns QMetry, acquired in May 2025, and presents it on the Zephyr product page as its standalone test management option. A shortlist of "Zephyr versus QMetry" is a choice between two products from one vendor.

The bill follows Jira, not your QA team#

This is the section that decides most evaluations, and it applies to every Zephyr edition sold through Atlassian.

Atlassian's Marketplace licensing terms tell buyers to "purchase the license tier that matches the number of users you have licensed for your host product", and then spell out the consequence: an app "will only function if its license matches or exceeds the tier of the host product", and that holds "even if only some of your licensed users need to use the app." There is no tester-only Zephyr licence. If your company holds 200 Jira licences, you buy Zephyr for 200 people.

Here is what that looks like in numbers. Zephyr prices were read from the Atlassian Marketplace pricing calculator on 1 October 2026 for Jira Cloud. QAM Hub prices are from our pricing page on the same day. The scenario is a team of 8 testers plus 4 people who only read reports.

Read the rows from the bottom. At 200 Jira users, Zephyr Standard costs five times what eight QAM Hub Standard seats cost, and the gap grows with every Jira user your company adds, whether or not they ever open a test. At 50 Jira users the two are in the same range, and on annual billing QAM Hub Standard for eight testers comes to $2,304 a year against Zephyr Standard's $3,405. At 10 Jira users Zephyr is far cheaper, and if that is your situation the licensing argument on this page does not apply to you.

A rule of thumb from the calculator figures, which is our arithmetic rather than anything SmartBear publishes: in the 11 to 100 user range Zephyr Standard averaged $6.81 per Jira user per month, and QAM Hub Standard is $27 per tester per month on monthly billing. QAM Hub costs less once your Jira instance is roughly four times the size of your QA team. Above 100 users Zephyr's per-user average falls (the calculator showed $5.57 at 200), so the crossover moves to about five times. Most organisations past fifty people are well beyond either ratio, because Jira spreads to product, support and management long before testing does.

Two Atlassian billing details make this sharper. Monthly cloud subscriptions are billed on peak usage: Atlassian's support documentation says "Your bill is based on the maximum amount of users you have throughout your billing period", and Marketplace apps are named in the list of products it covers, with the rollout starting on 14 July 2025. A two-week contractor ramp-up moves the Zephyr line on the invoice even though nobody new tested anything. And moving from Standard to Advanced, for example to get cross-browser automation, means upgrading the whole instance, not the QA team.

The same arithmetic applies to Xray and every other Marketplace test app, which is why switching between Jira apps does not change the bill's basis. Our QAM Hub vs Xray comparison works through the same licensing terms from Xray's side.

Zephyr executes tests, QAM Hub records the tests you already run#

This is the real product difference, and it is easy to miss because both tools say "automation" on the box.

Zephyr Standard and Advanced include SmartBear's no-code test automation. You record a web flow or describe it in natural language, Zephyr runs it in SmartBear's cloud, and the result lands against the test case in Jira. SmartBear's product page describes AI-assisted step suggestions, auto-generated validations and self-healing locators under its Halo AI branding. Advanced adds cross-browser runs on Chrome, Firefox, Edge and Safari (Standard runs on Chrome), three parallel runs, and email and SMS steps for flows such as sign-up verification. For a team with manual testers and no automation engineers, that is a real way to get regression coverage without writing code.

It comes with a published allowance. Zephyr's documentation lists monthly no-code automation executions by Jira tier, and states that "Unused automation test executions from the previous month will not roll over":

The same page gives a 30-day retention policy for automation execution data on both editions and marks SAML/SSO and static IP as unavailable for the no-code runner. It does not define what one execution is, and the difference matters a lot. If one execution is one test run, a 60-test smoke suite on 20 working days uses 1,200 a month, which is the full Advanced allowance on a 200-user instance and four times the Standard one. If it is one suite run, the allowance is generous. Ask SmartBear to confirm in writing before you size a pipeline around it. Note also that the allowance grows with your Jira user count, not with your QA team, so the same licensing logic governs how much you can automate.

QAM Hub does not run tests in a vendor cloud. It takes results from the frameworks you run in your own CI: an official Playwright reporter, an official Cypress reporter, and a command-line uploader for JUnit XML that covers WebdriverIO, Selenium, pytest, Jest, Vitest, TestNG, JUnit 5, PHPUnit, RSpec, Robot Framework, Newman and .NET. There is no monthly result count. Playwright traces, video and screenshots attach to each result, and a test linked to a manual case with TC-<number> in its title updates that case's coverage. Videos and traces are removed automatically after 30 days on Standard and 60 on Advanced. Screenshots and manual attachments are never removed automatically.

Where QAM Hub goes further is the history across runs. Test Explorer, on Standard and Advanced, tracks every automated test over the last 5 to 100 runs or a date range: pass rate, a stability score that counts only first-attempt passes, flakiness rate, duration, regressions and chronic failures. A Project Admin can quarantine a test with a reason. It stays visible and keeps running, and drops out of the health metrics until it is fixed. Our flaky test quarantine policy article covers how to run that well. We did not find a stability-history or quarantine feature in Zephyr's edition comparison or its Marketplace feature list.

QAM Hub also has an execution feature, Agent Flows (sold as AI Testing Agents), which runs test cases in Chromium, Firefox and WebKit through a runner you start on your own machine or CI. It is in beta on the Advanced plan, so weigh it as that. If built-in, production-grade no-code execution is the main thing you need today, Zephyr Advanced is the more finished product for that job.

AI in both products#

On Zephyr, the AI line runs between editions. Zephyr Essential has none of it. Zephyr Standard and Advanced include Zephyr Skills for Rovo, which the Marketplace listing describes as bulk-generating test cases from a Jira issue or a prompt and building a Zephyr agent for Rovo to "flag release risk, close coverage gaps, and alert your team". The same editions get smart test-step suggestions, generative AI that turns manual tests into automated ones, and SmartBear MCP, which lets Claude, VS Code, Microsoft Copilot and other assistants query Zephyr data. SmartBear runs a hosted MCP server for Zephyr with OAuth sign-in, and the local server connects to the Zephyr API with a token.

We found no published figures for how much of Rovo or Halo AI a Zephyr licence includes, beyond the no-code execution allowance above. Rovo is Atlassian's AI layer, so part of the experience depends on your Atlassian plan rather than on Zephyr. BearQ, SmartBear's autonomous testing agents, is a separate product: SmartBear's own FAQ of 10 April 2026 describes it as an early-access programme, and no pricing is published.

On QAM Hub, AI features are included in every plan. Every plan gets test case and checklist generation from a description or a screenshot, AI bug analysis and the in-app assistant. Standard adds AI fix suggestions for failing cases, automation run analysis that triages each failure as a product bug, a test bug or an environment problem, and run retrospectives. Advanced adds Release Readiness, Quality and Coverage Analytics (duplicates, unclear steps, an A to D score with evidence per finding), a knowledge base for AI context in beta and the AI Testing Agents in beta. In every case the AI proposes and a QA engineer accepts or rejects. MCP is on every plan: an agent connected through it acts as the token owner, a Viewer's token stays read-only, and every change is written to the audit log.

Feature comparison#

QAM Hub facts from the product and the pricing page on 1 October 2026. Zephyr facts from SmartBear's Zephyr documentation and product page and the two Atlassian Marketplace listings, all read on 1 October 2026.

What the ratings say, and which product they belong to#

Zephyr's ratings are scattered across old and new names, and several published comparisons quote one product's score for another. Here is each figure with its product, sample size and the date we read it.

The widely quoted "Zephyr 4.1 on G2 across 80 reviews" belongs to Zephyr Enterprise, the standalone product this page does not cover. The G2 profiles for Scale and Squad hold two and three reviews, which is not a sample. Zephyr's buyers review it where they buy it, in the Atlassian Marketplace, and there the evidence is substantial: 493 reviews for Zephyr and 900 for Essential. Those are the numbers to use.

Where Zephyr wins#

Price for small Jira instances. Up to 10 Jira users, $10 a month for Zephyr Standard is very hard to beat, and if your Jira headcount is close to your tester headcount the licensing argument above does not touch you.

No-code automation in the same tool. Record a flow, run it on a schedule, see the result on the test case in Jira, with cross-browser runs on Advanced. For a manual-heavy team with no automation engineers, that is a faster path to regression coverage than adopting a framework.

Cross-project reporting on every edition. Even Essential includes cross-space reporting and traceability. In QAM Hub cross-project reports and Projects Health start at Standard.

Everything stays in Jira. Jira permissions, Jira SSO, Jira search, and nothing new for a security team to review. The Rovo integration means a company already using Atlassian's AI gets Zephyr in the assistant its people already open.

A large review base. Close to 1,400 Marketplace reviews across the two listings against our zero.

Where QAM Hub wins#

You pay for testers. Eight testers cost eight seats and bring eight free viewer seats, and growing the Jira instance does not change the bill. That is most of the case for a mid-sized or larger company.

Your automation stays yours. Results come from the frameworks your engineers already maintain, with no monthly result count, and the Playwright trace, video and screenshots sit on the result the next morning. Nothing about how often you run tests depends on your Jira tier.

Flakiness is a managed object. Stability across runs, a quarantine list with reasons, regressions and chronic failures, which is the information a team needs before it can trust a red build. The mapping automated test results to test cases article explains how results connect to the case record so coverage stays accurate.

Trackers beyond Jira. If part of your engineering works in GitHub Issues, GitLab or Linear, QAM Hub connects to all seven supported trackers from one place. Requirements come in from your tracker, and each linked issue gets a QAM Hub comment listing the test cases that cover it, their latest results and the bugs found.

Your test assets do not depend on your Jira licence. QAM Hub integrates with Jira; it does not live inside it. If your company changes Jira plans, merges instances or leaves Atlassian, the test library is unaffected.

Leaving Zephyr: what moves and what to plan for#

QAM Hub has import wizards for TestRail and Qase. There is no Zephyr wizard today, so the route is CSV. Export the case library from Zephyr, map the columns to QAM Hub's case fields in the CSV import, and check the result on one project before running the rest. The CSV import and export guide covers the field mapping. Requirements and checklists come in the same way. On the Advanced plan, QA Madness runs the import and the field mapping together with you.

Plan for three things. First, execution history: decide whether past Zephyr cycles need to move at all, or whether keeping read access to the old Jira data as an archive is enough. Moving them means scripting against Zephyr's REST API, and the effort depends on how many cycles you keep and how you want them to look on the other side. Second, attachments: check whether your export carries images and files, and if it does not, move the cases that depend on screenshots by hand or through the API. Third, automation: you do not need to migrate old automated results. Point your Playwright, Cypress or JUnit XML pipeline at QAM Hub and history starts building from the first run. Tests created in Zephyr's no-code recorder do not transfer as code. They need to be rebuilt in a framework or kept in Zephyr.

Running both tools in parallel for one release is the safest way to switch: the library and new results in QAM Hub, the old cycles readable in Jira. If you are still deciding whether to leave Jira-native test management at all, test management tools for Jira covers the decision from the start, and Xray alternatives covers the wider field of Jira apps.

When you should not switch#

Stay on Zephyr if your Jira user count is close to your tester count, because then the price is low and you get Jira-native workflow with it. Stay if your regression coverage is built on Zephyr's no-code recorder, it fits inside your execution allowance, and nobody on the team wants to maintain a framework. Stay if a new SaaS vendor review would cost your security team more than the licence difference, which in regulated companies can decide the question by itself.

Switch when a specific cost shows up: Jira seats paid for people who never test, automation runs rationed by an allowance, a flaky suite nobody measures, or engineers working in trackers other than Jira. If none of those applies, keep what works.

How we compared these tools#

Zephyr's editions, features and limits were read from SmartBear's Zephyr documentation and product page and from the two Atlassian Marketplace listings on 1 October 2026. Prices come from the Marketplace pricing calculator for Jira Cloud on the same day, monthly billing unless stated. Licensing and billing rules come from Atlassian's own licensing terms and support documentation. Ratings are quoted with their review counts, their product and the date they were read. The verification date at the top of this page is the date the facts were last checked, and these pages are re-verified quarterly.

QAM Hub is built by QA Madness, so read our ranking of ourselves with that in mind and check the working above: every Zephyr fact on this page comes from a page you can open yourself.