Safari/WebKit coverage across AI test tools

Most AI test platforms run your tests in one engine: Chromium. If your users are on iPhones, that means the browser your revenue walks through is the one your test suite never opens.

This page is a factual engine-coverage comparison across AI/natural-language test automation platforms. Every claim about another vendor links to that vendor’s own documentation, with the date we read it. If you spot something outdated, tell us at support@marriska.com and we will correct it.

Why Safari coverage is not optional

• On iOS, real-world browsing overwhelmingly runs on WebKit — for most of the world, every browser on an iPhone (including Chrome) is WebKit underneath.
• WebKit renders and behaves differently: flexbox and gap quirks, date and file inputs, sticky positioning, autoplay and permission policies, Safari’s Intelligent Tracking Prevention around cookies and logins.
• The practical consequence: a checkout that passes 1,000 Chromium runs can still be broken for every iPhone user — and nothing in a Chromium-only suite will ever tell you.

Firefox is the same argument at smaller scale: a real Gecko engine, real differences, usually zero coverage.

Engine coverage, from each vendor’s own docs

Readings dated 2026-08-13.

Platform	Chromium	Firefox	WebKit (Safari)	Hosted runs
Marriska	✅	✅	✅	✅ included, from $0
Momentic	✅	❌	❌	⚠️ deprecated
Meticulous	✅ (Chrome)	❌	❌	runs in your CI
Autosana	✅	✅	❌	✅

Sources, verbatim from the vendors:

• Momentic — the browsers page documents Chromium-family targets only; no Firefox or WebKit appears anywhere in their docs index. Hosted cloud runs are deprecated in favour of running in your own environment.
• Meticulous — their architecture overview states tests run in Chrome. (Different model entirely: replay of recorded sessions, no authored tests.)
• Autosana — the web testing matrix lists Chrome, Chromium, Edge and Firefox. Three of those are the same engine (Blink); WebKit does not appear.

Several other platforms in this category — including vendors we otherwise rate highly — publish no browser-engine documentation at all: their public docs and pricing pages do not state which engine executes your tests. We don’t guess, so they are not in the table. If you’re evaluating one, it is a fair first question to ask them.

What about Playwright itself?

Playwright — the open-source framework Marriska executes on — supports all three engines and is free. If you are happy writing and maintaining test code and running your own infrastructure, it is an excellent choice, and its test agents can help author specs. Marriska is for the other case: tests written in plain language, stored and versioned on a shared platform, executed on hosted infrastructure across all three engines — with no runners to maintain.

How it works in Marriska

Cross-browser is not a tier or an add-on. Every plan, including Free, runs the same stored test on any combination of Chromium, Firefox and WebKit:

1. Write the test once, in plain language — it becomes explicit, editable steps.
2. Pick browsers per run or per schedule: one engine, two, or all three.
3. Each browser gets its own step results, screenshots, and visual-regression baselines — WebKit is compared against a WebKit baseline, never a Chromium one.

See Browser support for the engine matrix and per-browser behaviour, or start on the Free plan — 300 hosted runs a month, all three engines, no card.

Methodology & fairness

Competitor information on this page is limited to what each vendor publishes in their own documentation, read on the date stated. Absence of documentation is reported as exactly that — not as a claim that a capability is missing. Vendors: if your engine support has changed, email support@marriska.com with a doc link and we’ll update this page.