# 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 step by step, checks what actually happened on screen, and returns a verdict with named checks, severity-rated findings, screenshots and a session video.

ZeusQA differs from Playwright, Cypress and Selenium: those run scripts you write and maintain, while ZeusQA writes and adapts the steps itself from your description, reads the screen like a user, and reports console errors, failed network requests and UI problems a fixed script would miss.

## Key pages
- [Home](https://zeusqa.ai/): product overview, live browser view, session video, evidence-based reports.
- [Pricing](https://zeusqa.ai/pricing): Free ($0), Pro ($39/mo), Team ($149/mo).
- [Machine-readable pricing](https://zeusqa.ai/pricing.md)
- [Sign up](https://zeusqa.ai/signup): free, no invite code, no card.

## Use cases
- [Sign-up flow testing](https://zeusqa.ai/use-cases/signup-flow-testing)
- [Cart and checkout testing](https://zeusqa.ai/use-cases/checkout-testing)
- [Responsive testing](https://zeusqa.ai/use-cases/responsive-testing) at 1440x900, 834x1060 and 430x790
- [Smoke testing](https://zeusqa.ai/use-cases/smoke-testing)
- [Accessibility testing](https://zeusqa.ai/use-cases/accessibility-testing): ARIA tree and keyboard, not a WCAG audit
- [Form validation testing](https://zeusqa.ai/use-cases/form-validation-testing)

## Free tools (no signup, no account, run in the browser)
- [Viewport size reference](https://zeusqa.ai/tools/viewport-size): live viewport readouts plus 100+ device CSS viewports from Playwright's descriptors.
- [Selector brittleness checker](https://zeusqa.ai/tools/selector-check): scores CSS/XPath selectors and explains why each will break.
- [Test prompt builder](https://zeusqa.ai/tools/test-prompt-builder): composes a browser-agent test prompt, a generic agent prompt, or a manual test plan.

## How it works
1. Describe the test in plain English (a URL and what to check).
2. A frontier model drives a real browser, acting in small steps and looking at each screenshot before the next action.
3. You get a verdict — pass, fail or blocked — with checks, findings, the exact scripts and the screenshots and video that back them.

## Models
Any tool-calling model with vision available through LLM Gateway: GPT-6 Astra, Claude Sonnet 4.6, Gemini 3 Pro, GPT-5.6 Luna, Claude Haiku 4.5, DeepSeek V4, and 270+ others. Choose speed or depth per run.

## Plans
- Free: $0 — 5 test runs total, 1 project, live browser view.
- Pro: $39/month — 300 test runs/month, unlimited projects, live view + session video, priority models.
- Team: $149/month — 1,500 test runs/month, organizations with members and roles, shared projects.

## Devices
Runs happen at three visible viewport sizes, not screen sizes: laptop 1440x900, tablet 834x1060 (touch, iPad user agent), phone 430x790 (touch, mobile user agent). Portrait or landscape.

## Limits (things ZeusQA does not do)
- No scheduling, cron, uptime monitoring or CI trigger. A run is started by a person.
- Chromium only. No Safari or Firefox.
- No visual regression or pixel diffing against a baseline.
- No WCAG conformance report and no axe-core run.
- No email inbox access, so it cannot click a verification link.
- No CAPTCHA solving. A challenge usually ends a run as "blocked".
- Runs are private to your account; there are no public shareable run links.