You’re already paying
for an AI subscription.
Get the full QA loop for free.

Every week someone asks the same question: “What’s the catch on the free tier?”

There isn’t one. The free tier is the product, not a trial. If you already pay for Claude Code or Codex, you have everything you need to run a full QA loop — crawl, generate, run, fix — against your own app, without leaving your terminal and without spending another cent on models.

This post is the boring, useful version: exactly what you get, exactly what it costs, and exactly how to wire it up.

What’s in the Free Tier

The three things people consistently miss when they first look at the pricing page:

• “Unlimited local execution” is the headline. The crawl budget is the only daily metered resource. Once the crawl has produced your tests, you can re-run them locally as often as you want — in CI, on every commit, in a watch loop — and nothing on QualityMax’s side counts that against you.
• You also get 60 cloud-sandbox minutes a month. Free. Real isolated cloud execution against any URL — useful for the runs you’d rather not pin to a developer laptop: nightly checks, on-merge smoke suites, or tests against staging when your CI runner can’t reach it.
• qmax-code is on the free tier. Not the “trial”, not the “limited preview” — the same binary that paying customers use, with the same MCP toolset and the same PR bot.

The Trick: Use the Subscription You Already Have

qmax-code is BYOLLM. It talks to whichever model provider you point it at. If you already have a Claude Code subscription or a Codex / ChatGPT Plus subscription, you do not need another model bill on top.

The way the pieces fit together:

• Your CC or Codex subscription handles the heavy lifting — planning, file edits, agentic loops — using your Anthropic or OpenAI quota.
• qmax-code sits alongside CC/Codex in the same terminal as an MCP server. CC and Codex call into it whenever they need to crawl your app, generate tests, run them, or read a session record.
• The QualityMax platform is what stores the test suite, the run history, and the PR bot output. That part is on us, on the free tier.

No second LLM bill. No double-paying for tokens. The model spend you’re already committed to becomes your full QA stack as a side effect.

What it looks like in a real session

$ claude   # or: codex

CC> Crawl https://staging.myapp.dev with qmax-code and write
    Playwright tests for the signup flow

qmax> starting AI crawl — project: myapp / staging
qmax> 12 pages discovered · 4 forms · 1 auth flow
qmax> generated: tests/signup.spec.ts (37 lines, 3 assertions)
qmax> 1 of 50 free-tier crawls used this month

CC> run the test locally and show me the result

qmax> spawning local browser…
      ✓ signup flow — valid email          (1.4s)
      ✓ signup flow — duplicate email      (1.1s)
      ✗ signup flow — weak password       (0.9s)
       expected validation message, got blank toast

CC> fix the failing case

CC  edited frontend/signup.tsx — added validation handler
qmax> retesting…
      ✓ 3/3 passed — local execution, no cloud minutes consumed

That round trip — crawl → generate → run → fix → retest — happened in your terminal, using one of your ten monthly crawls and zero metered execution minutes. The model tokens came out of the Claude / OpenAI subscription you were going to pay for anyway.

Stay in Your Own Ecosystem

The reason this design matters: you don’t have to leave the tools you already work in.

The generated tests are normal .spec.ts files. They live in your repo, run in your CI, get reviewed in your PRs. QualityMax is the engine that produces and maintains them — not a walled garden you have to migrate into.

Fast Feedback Without a Round Trip

The unlimited-local-execution piece is what makes the loop feel different from cloud-only QA tools. Most platforms meter every run because every run lives on their infrastructure. We don’t, because most of your runs shouldn’t live on ours.

• Pre-commit feedback in seconds. The local agent boots a real Chromium and runs your generated suite directly against localhost:3000. You don’t have to deploy anywhere first.
• Watch-mode reruns are free. Iterate on a flaky selector or a new feature without watching a meter tick up. CI is where billing matters — local development should be friction-free.
• The cloud is for what the cloud is good at. Use the AI crawl quota when you want a fresh discovery pass or a regenerated suite against a deployed environment. Spend your 60 free sandbox minutes on the runs you don’t want pinned to a laptop. Use local execution for everything else.

Where the platform earns its keep

Even on the free tier, the QualityMax side of the loop holds the parts you don’t want on your laptop:

• The PR bot. Connect a GitHub repo and every PR gets reviewed for security gaps and tech debt automatically. Findings land as PR comments with severity, file path, and a one-line fix.
• Test-run history. Pass/fail timelines, screenshots, and traces for every run you push to the platform — so when a test starts flaking three weeks from now, you can see exactly what changed.
• AI crawl & regeneration. The expensive part — the multi-model crawl pipeline — runs in our cloud so your laptop fan stays quiet.
• Cloud Sandbox minutes. 60 every month, free. Isolated browsers in our EU sandbox for the runs that shouldn’t live on a developer machine — scheduled checks, on-merge smoke suites, or CI jobs whose runners can’t reach your staging URL.

Your Code Stays Where It Belongs

One of the nicest side effects of the BYOLLM design: your source code never has to leave your machine to participate in the loop.

Your code stays on your laptop, your model bills stay on your account, and the artifacts that do live with us are the ones you’d want to keep — until the day you decide otherwise.

The Three-Minute Setup

That’s the entire onboarding. No “14-day trial” counter, no credit card on file, no feature flag waiting to expire. The free tier was designed to answer one question: can a developer who already pays Anthropic or OpenAI for an AI coding assistant get a serious QA stack without paying for tokens twice? — not to be a teaser for the paid one.

When you outgrow it — more than 50 crawls a month, more than 5 projects, more than 60 cloud-sandbox minutes, or team seats — the Starter plan is €79/month. Until then, the loop is yours to run as often as your laptop can keep up.

curl -sL https://qualitymax.io/static/install-qmax-code.txt | bash