BlogFlowQA
What AI test automation costs a small team: the tool, the usage and the hours
AI test automation has three costs: the plan, the usage or credits past it, and the hours someone spends reading failed runs and approving repairs. What a team of one to five pays at 30, 100 and 300 tests, from published prices checked on 18 November 2026.
AI test automation costs a small team between $0 and $900 a month to run 30 tests once a day on a self-serve tool, and $2,500 a month on the one managed service that publishes its price. The bill has three parts: the plan, the usage or credits past what the plan includes, and the hours someone on your team spends reading failed runs and approving repairs. For a suite of 30 tests run once a day, the first two parts come to $0 on FlowQA's Free plan, $45 to $135 on Stably, $500 on Momentic and $900 on QA Wolf.
TestFinch makes FlowQA. Every figure for another company comes from its own pricing page, checked on 18 November 2026.
What are the three costs of AI test automation?#
AI test automation means software that writes browser tests for your app and runs them after you change something, so you learn that checkout broke before a customer does. The set of tests you keep running is called a regression suite.
The tool. A monthly plan, priced by what it includes: people, projects, saved tests and a number of test runs.
The usage. A test run is one test played from start to finish, once. Every tool meters something past the plan: runs, steps, browser minutes, AI tokens or credits. A suite that runs daily uses 30 times what the same suite uses run monthly.
The hours. A failed run is either a bug in your app or a test that no longer matches the page. Someone has to open it and decide which. No pricing page lists this cost.
What makes the tool cost more?#
- People. FlowQA's Free plan is for 1 person and Starter allows up to 5. QA.tech's Starter allows up to 5 users. TestGrid charges $249 for each seat. Stably and QA Wolf charge nothing per person.
- Apps. FlowQA counts projects: 1 on Free and 3 on Starter. QA.tech's Starter covers 2 applications.
- Saved tests. FlowQA's Free plan keeps up to 100 tests, and paid plans have no limit. Checksum prices only by the number of workflows it maintains.
- Who does the work. A plan that includes an engineer costs more than one that does not: Bug0 is $2,500 a month.
What makes usage cost more?#
Usage is tests × runs a day × the browsers and screen sizes each run covers.
Running more often. Running after every deploy instead of once a night multiplies the bill by your deploys per day on any tool that charges per run or per step.
Longer tests. A per-step tool charges for length: at Momentic's $0.0125 a step, a 10-step test costs 12.5 cents a run and a 40-step test 50 cents. A per-run tool charges the same for both, up to a time limit. QA Wolf's $1 covers 5 minutes, and a 7-minute test costs $2.
More screens. On FlowQA each browser and device a test runs on counts as one cloud test run, so a suite run at three screen sizes uses three times the allowance.
AI generation and repairs are metered too. Momentic charges $0.025 for an AI-generated or failure-recovery step, and $1.25 for each triage or auto-healing call. Stably charges AI tokens at $0.30 to $15 per million, depending on the model. On FlowQA an accepted AI-generated test past the plan's allowance costs 10 credits and a test repair costs 5 credits, at 1 cent a credit. Runs on your own computer through FlowQA Desktop are free on every plan.
What should a team of one to five expect to pay?#
An illustration with round numbers, priced from the vendors' pages on 18 November 2026; it is not a customer's bill.
Each suite in the table runs once a day for 30 days, in one browser, with tests of about 10 steps that take about two minutes.
| Suite | Runs a month | FlowQA, one person | FlowQA, five people | Stably | Momentic | QA Wolf |
|---|---|---|---|---|---|---|
| 30 tests | 900 | Free | Starter, $49 | $45 to $135 | $500 | $900 |
| 100 tests | 3,000 | Free plus 2,000 credits | Starter, $49 | $150 to $450 | $500 | $3,000 |
| 300 tests | 9,000 | Starter, $49, plus 6,000 credits | Starter, $49, plus 6,000 credits | $450 to $1,350 | $1,215 | $9,000 |
How each column is worked out:
- FlowQA is the plan the pricing page's estimator picks. Free includes 1,000 cloud test runs a month and Starter 3,000. A run past the allowance costs 1 credit, and 1,000 credits cost $10. For one person with 100 tests it picks Free with credits over Starter, and at 300 tests Starter with credits over Team at $149, because the credits cost less than the next plan.
- Stably uses its page's own example, a typical run at about $0.05 to $0.15 of credits, times the runs. Its plans include $10, $60 or $250 of credits a month.
- Momentic is steps × $0.0125 plus browser minutes × $0.005. For 30 tests that is 9,000 × $0.0125 + 1,800 × $0.005 = $121.50, and for 100 tests $405. Both are past the Free plan's $50, which cannot be topped up, and inside the $500 Scale plan. For 300 tests it is 90,000 × $0.0125 + 18,000 × $0.005 = $1,215: Scale plus $715 of added funds.
- QA Wolf is $1 a web run.
A flat fee does not move with any of this: Bug0's $2,500 covers unlimited runs for up to 500 user flows. AI testing tool pricing compared has the same arithmetic for 15 tools, including the ones whose credits cannot be priced from what they publish.
What do the hours cost?#
No vendor publishes a figure you can budget from, so measure your own for a month. Count the failed runs, and the minutes each took to sort into "the app is broken" or "the test is out of date".
The number depends on:
- How often the app's pages change. A renamed button fails every test that taps it.
- What a failure shows you. FlowQA shows the failed step with its screenshot. What a failing step looks like has an example.
- Who makes the repair. With plain Playwright, a developer edits the test. FlowQA proposes one small change, replays it twice, and changes the test only once a person approves it. Stably's homepage says it "proposes updates for test drift" for review.
A flaky test fails on some runs with no bug and no page change, and each of those failures is one more to read. Flaky tests: causes and triage covers how to find and remove them.
When is a free plan or open source enough?#
A free plan is enough while one person looks after one app and the suite runs about once a day. As of 18 November 2026, these free plans are published:
- FlowQA Free: 1 project, 100 tests, 1,000 cloud test runs and 10 AI-generated tests a month.
- Momentic Free: $50 of usage a month, which is 4,000 steps before browser minutes.
- Stably Hobby: $10 of credits a month and one browser at a time.
- TestSprite Free: 150 credits a month and one environment.
- Functionize Studio Free: 800,000 credits a month with no time limit.
Open source is the right answer when a developer on the team wants to write and own the tests as code. Playwright is published under the Apache-2.0 license, so there is nothing to buy. You pay in the time to write each test, the minutes your CI spends running them, and the time to fix them when the app changes. Playwright vs record-and-replay covers when hand-written tests are worth that time.
If you built the app with an AI coding tool and have never written a test, start with a free plan and a short list of what must not break. How to test a vibe-coded app before launch has that list.
When is paying someone to own the tests the right call?#
Pay for a managed service when nobody on the team will open a failed run within a day.
As of 18 November 2026, the published ways to buy that:
- Bug0: $2,500 a month, month to month, with an engineer, unlimited runs and up to 500 user flows.
- QA Wolf: its managed services are quoted by sales, based on the tests you need.
- TestFinch: our team sets FlowQA up for you for $2,500, one-off, and looks after it for $1,200 a month. Setup covers up to 30 journeys, built in your own FlowQA workspace. Care covers up to 50 tests, with a QA person who looks at every failed run, files the bugs with evidence and approves repairs. The service is in beta and covers web apps only.
Setup without care suits a team that can read failures but has no time to build the first suite.
How do you budget before you buy?#
- List the journeys that must not break: sign-up, sign-in, checkout, the one screen your customers pay for. The regression test template is a starting list.
- Multiply tests × runs a day × screen sizes to get runs a month.
- Price that number on each tool's page in its own unit, and put FlowQA's through the estimator.
- Start on a free plan and count the failed runs you read in the first month.
- Buy the plan that fits the runs you measured.
FlowQA explores your site, writes the tests, keeps each one only once it fails on a page broken the way its checks describe, and runs them after every change.
See FlowQA