Describe the test.
A real browser runs it.
ZeusQA turns a URL and a plain-English test into a real run. A frontier model drives Chromium step by step, checks what actually happened on screen, and hands you a verdict with checks, findings, screenshots and video.
No credit card. 5 free tests. Have an invite code?
The problem
Scripts click. They don't look.
A recorded test passes while the page is broken, because it never checked what a person would see.
Selectors rot
Rename a button and a hundred scripts go red. A model reads the screen the way your users do and finds it anyway.
Console errors hide
The UI looks fine while the console fills with exceptions and 404s. Every run collects them as findings.
Nobody reads the artifacts
Screenshots pile up in CI. Here the model looks at each one before its next step, and the report shows you the ones that matter.
Everything ZeusQA does
One prompt, real evidence
Each capability, highlighted.
Prompt-driven testing
Write the test in plain English. No selectors, no framework. The model plans the steps and adapts when the UI changes.
Any model
GPT-6 Astra, Claude, Gemini or 270+ models through one gateway. Pick speed or depth per run.
Live browser view
Watch the real Chromium as the model drives it, at laptop, tablet or phone size. Stop a run at any point.
Session video + evidence
Every run records a video and a verdict: named checks, severity-rated findings, the exact scripts and the screenshots that back them.
Findings a script would miss
Console errors, failed requests and copy that contradicts behaviour — surfaced as findings, not buried in logs.
Organizations & projects
Group runs by project, share them across your team, and manage roles and billing per organization.
How it works
Three steps, one report
- 1
Write the test in English
A URL and what you want checked. Validation rules, a happy path, an edge case.
- 2
Watch the browser work
The model looks first, acts in small scripts, then looks again. Follow every step live.
- 3
Read the verdict
Pass, fail or blocked, with named checks, severity-rated findings and the evidence behind each one.
Pricing
Start free. Scale when you ship.
Every plan runs real browsers driven by frontier models. No credit card to start.
Free
Kick the tires.
- ✓5 test runs (total)
- ✓1 project
- ✓Live browser view
- ✓Community support
Pro
Most popularFor builders shipping fast.
- ✓300 test runs / month
- ✓Unlimited projects
- ✓Live view + session video
- ✓Priority models (Astra)
- ✓Email support
Team
For teams that test together.
- ✓1,500 test runs / month
- ✓Organizations with members
- ✓Roles & shared projects
- ✓Everything in Pro
- ✓Priority support
Prices in USD. During the beta, checkout runs in Stripe test mode.
A real run
Sign-up flow on llmgateway.io, tested by Luna
Six scripts: it found the Get Started link, submitted the empty form, tried a bad email and a short password, created a throwaway account, then checked where the session landed and whether the console was clean.
The finding a script would have missed
The toast said check your email to verify before signing in, yet the session was redirected to onboarding with a working API key. Verification isn't enforced and the copy contradicts the behaviour — with the screenshot to prove it.
FAQ
Questions, answered
- What is ZeusQA?
- ZeusQA is prompt-driven QA for web apps. You give it a URL and a plain-English test; a frontier model drives a real Chromium browser, checks what happened on screen, and returns a verdict with checks, findings, screenshots and a session video.
- How is it different from Playwright or Cypress?
- Those run scripts you write and maintain. ZeusQA writes and adapts the steps itself from your description, reads the screen like a user, and reports console errors, failed requests and UI problems a fixed script would miss.
- Which AI models can I use?
- Any tool-calling model with vision on LLM Gateway — GPT-6 Astra, Claude, Gemini and 270+ others. Pick speed (Luna, Haiku) or depth (Astra) per run.
- How much does it cost?
- Free includes 5 test runs. Pro is $39/month for 300 runs, unlimited projects, live view and video. Team is $149/month for 1,500 runs and organizations with members.
- Do I need to verify my email?
- You can sign up and run free tests immediately. Email verification is required before subscribing to a paid plan.
- Is it safe to run on my site?
- The model treats page content as untrusted and avoids destructive actions unless your prompt allows them. It drives a real browser, so use staging or throwaway accounts for anything that writes data.
Your first test takes about a minute
Paste a URL, describe the flow, watch the browser do the rest.