QA Reef vs every competitor
One table, QA Reef against all 19 platforms we researched, on the capabilities that actually separate them. Every competitor cell is drawn from that vendor's own public pages, as of August 2026.
Where a vendor does not publish a price, the cell reads "Not published" rather than a guess. QA Reef is pre-launch with no customers. It is not affiliated with any product named here; the names identify the products being compared, nothing more.
The feature matrix
The QA Reef column is highlighted. Scroll the table sideways for the rest. Each platform name links to its full, sourced write-up.
| Capability | QA Reef | QA Wolf | MuukTest | Rainforest QA | Testim | mabl | Autify | Functionize | Reflect | BrowserStack | Checkly | Applitools | testRigor | Virtuoso QA | Katalon | Testsigma | Momentic | Meticulous | outsourced QA agencies | your own Playwright setup |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Open-source core | Open coreCLI Apache-2.0, core MIT, repo in progress | PartialCLI and flows OSS, platform closed | NoCI helpers MIT | NoCLI is MIT | Noexamples only | NoMIT SDKs and CLI | NoMIT SDKs and orbs | No | No | NoSDKs | PartialCLI, SDK, Helm Apache-2.0, shipping | NoApache-2.0 SDKs | Noplugins only | No | Nofree EULA, not OSS | YesApache-2.0 core | NoMIT CLI | NoISC SDK | N/Abuild on OSS frameworks | YesApache-2.0 |
| Self-hostable | Yesyour hardware or ours | Nohosted grid | Not stated | Nohosted on GCP | Not stated | Nomanaged cloud | PartialNexus on-prem exec | Nocloud ML required | Nocloud browsers | PartialAutomate self-hosted | Partialprivate locations | Not published | Yeson-prem option | Unclearcloud platform | Yeson-prem | YesOSS or on-prem | Nolocal, CI, or cloud run | Not published | Usuallyyour CI | Yesentirely yours |
| Who writes the tests | Yourecord, or agent from a goal | Vendor QA team | Vendor engineers | Youno-code | Youlow-code | YouAI-generated | YouNL, low-code, or code | AIfrom your prompt | Youno-code | Youlow-code | YouPlaywright code | YouSDK or no-code | Youplain English | Younatural language | Youagent, no-code, or code | Youplain English | You plus agentNL spec | None writtenauto from real sessions | Vendor engineers | Youcode or codegen |
| Pricing model | Flat + usagehosted from $20/mo, $0.015/min past included | Usagecredits plus runner-minutes | Outcome-based | Sales quote | Free tier plus sales | Creditsquoted by sales | Published tiers | Published tiers | Credit tiersvia sales | Published plus sales | Published tiers | Published plus sales | Published plus sales | Consumptionvia sales | Per seatpublished | OSS free plus sales | Published tiers | Not publishedno public page | Sales quote | Freeself-run |
| Published entry price | $0 core; $20/mo hostedSolo, 1,333 min included | $0.01/credit + $0.15/minself-serve card | Not published | Not published | Free Communitypaid via sales | Not publishedfrom 500 credits/mo | $99/moCore, annual | Free, then $20/mo | Not published1 credit per web test | $29/moLive, annual | Free, then $24/mo | $667/moStarter, annual | Free, then $300/moPrivate | Not published | $84/seat/moannual | OSS freecloud not published | Free, then $125/mo | Not published | Not publishedno agency lists a rate | $0plus engineer time |
| Flaky-run handling | Heal ladderplus an UNMEASURED verdict | DOM analysis, then human verify | Human team plus smart locators | Fallback targetingscreenshot, DOM, AI | Smart locators | Auto-heal | Fix with AI | ML self-heal | Self-heal | Self-healing agent | N/Amonitoring, not testing | Deterministic plus Exec Cloud | AI, no XPath | ML self-heal | Locators, then LLM | Auto-heal locators | Self-heal specs | Flake-free scheduler | Humans absorb it | You build itretries, quarantine |
| Self-healing | Quarantinedmodel heals wait for a human | YesDOM-based | Yesplus human | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Nonot its job | YesExecution Cloud | Yes | Yes | Yesauto-applies | Yes | Yes | N/Ano locators | Manual | NoDIY |
| Legacy / canvas / iframe | YesOCR screen-reading | Not statedDOM locators | Not stated | Partialpixel-matching | Salesforcecanvas not stated | Not stated | Not stated | Noweb UI | Partialvisual detection | Not stated | N/A | Partialvisual | Broaddesktop, mainframe, 2FA | Packaged appsSalesforce, Workday | Desktopcanvas no | Salesforce, desktop | Not stated | Not stated | Yeshumans handle anything | Hardmanual coordinates |
| CI / deploy-check | CLI, Action, Vercel check | CLI plus CI | CI/CDCircleCI orb | CircleCI, GH Actions, CLI | Jenkins, CircleCI, GH | GH, GitLab, Jenkins | CircleCI, Jenkins, GH | None named | GH Actions, Jenkins, Azure | All major CI | As code via CLI | Manylist unverified | Jenkins, CircleCI, Azure | GH, Azure, Jenkins | Very broad | Jenkins, CircleCI, Azure | GH Actions, CircleCI | GH, GitLab, CircleCI | Your stack | Any pipeline |
| Website to API mode | Yesnetwork-capture | No | No | No | No | No(does API test type) | No | No | No | No | No(runs API monitors) | No | No | No | No | No | No | No | No | No |
| Data ownership | Yoursself-host or per-workspace | Their platform | Your scripts, their platform | HostedSOC 2 Type II | Hosted | HostedSOC 2 Type II | Hostedon-prem exec option | Hosted | Hosted | Hostedself-host exec option | Hostedplus private agents | Hosted | Hostedon-prem option | Hosted | Hostedon-prem option | Yours (OSS)or cloud | Their cloudSOC 2 Type 2 | Vendorrecords real sessions | Yours, plus contractor access | 100% yours |
| Agentic maintenance | Yesexplore and heal agent | AI plus human team | Human team | AI drafting | AI agent | AI-native | Aximo agent, Genesis | AI agents | AI | Self-healing agent | No | Autonomous product | AI | ML | Agent modeMCP | AI | Yesagents | Automaticno agent | Humans | No |
| Human-in-the-loop review | Yesheal review plus agent inbox | YesQA engineers | YesQA team | You steer | You | You | Youreview the AI fix | You | You | You | You | Youapprove visual diffs | You | You | You | You | You | Youapprove diffs | Yesthe vendor is the human | You |
| CAPTCHA stance | Handles iton apps you are authorized to test | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Not stated | Humans solve it | You handle it |
Two honest notes we keep on the record: on the open-source row, Testsigma and Checkly ship an OSI-licensed core today while QA Reef's is still being extracted to a public repo, and Momentic has SOC 2 and mobile that QA Reef does not. CAPTCHA handling is rarely documented by any vendor, so those cells show only what each states, not necessarily the full capability.
By category
Prefer prose to a grid? Each write-up states what the platform is good at and who should pick it over us.
Codeless AI platforms
Rainforest QA · Testim · mabl · Autify · Functionize · Reflect
Platforms & grids
Plain-English platforms
testRigor · Virtuoso QA · Katalon · Testsigma
AI test agents
People, not products
Comparing frameworks rather than vendors? We read the end-to-end suites of 16 public products and wrote up what a Cypress to Playwright migration actually costs, and what the same counts say about flaky tests. If the decision is still open, Playwright vs Cypress puts the two side by side on counts rather than feature lists.
What QA Reef actually is
An open-core QA agent you deploy like a Vercel project: record or describe a flow, get a deterministic Playwright spec, run it on your hardware or ours, and get PASS, FAIL, or UNMEASURED back.
- Open core. The
qareefCLI is Apache-2.0; the automation core underpackages/coreis MIT, with extraction to a public repo in progress. - One recorded flow, four run modes. Explore and record a flow, then run it for data entry, scraping, a website-to-API endpoint (real network capture), or a plain test. Deployed like a Vercel project over a GUID-addressed workflow → suite → flow → step tree.
- Deterministic codegen. The Playwright spec is written by code. The model only proposes a title and assertions, each stamped
decided_bywith the exact model id and its cost. - Self-healing that refuses. Model-proposed heals are quarantined for a human, never auto-promoted. A coordinate click, or a run that healed too much, returns UNMEASURED, not a pass.
- It reads the screen. Apple Vision OCR or a vision model, for canvas and legacy UIs with no stable locators. A free local-model backend means a self-hosted run can cost $0.
- Pre-launch. No customers, no case studies, no SOC 2, no native mobile, no managed human-QA service. Hosted plans are published but billing opens in waves. Every page here says so.
Questions
How honest is this matrix, really?
We sell one of the columns, so weigh it accordingly. What we hold to: every competitor cell is read from that vendor's own public pages, dated as of August 2026, and phrased as what they publish, not as an absolute claim. Nothing about pricing, customers or benchmarks is estimated. And we keep the cells where a rival beats us in plain sight, not buried.
Where does a rival beat QA Reef on this table?
On the open-source row, Testsigma and Checkly ship an OSI-licensed core today while QA Reef's is still being extracted to a public repo. Momentic has SOC 2 and native mobile that QA Reef does not. The managed services (QA Wolf, MuukTest, outsourced agencies) put a human in front of every failure, which QA Reef does not offer.
Why do so many cells say "Not published"?
Because those vendors do not publish the figure. Most enterprise QA pricing runs through a sales call. We would rather write "Not published" than repeat a number from a third-party aggregator as if it were the vendor's own. Eight of the platforms here do publish a real entry price, and those are quoted exactly.
How much does QA Reef cost?
Published on the pricing page. The open-source core is free, self-hosted: you pay only for your own model calls, each logged with its model id and cost, and with the free local-model backend that bill is $0. Hosted plans: a free tier with 500 browser-minutes a month, Solo at $20/mo with 1,333 minutes included, Team at $99/mo with 10,000, Scale at $499/mo with 60,000, all then $0.015 per browser-minute. Hosted billing is pre-launch and opens in waves with $20 of signup credit.
Is QA Reef affiliated with any of these companies?
No. Every product name here belongs to its owner and is used only to identify the product being compared. No logos are used, and nothing here implies a partnership, sponsorship or endorsement in either direction.
Going deeper on QA Wolf specifically?
Three longer pages: the alternative case, their sourced pricing, and the architecture side by side.
QA Wolf alternative QA Wolf pricing Architecture, in depth