How we test AI coding tools
Is the AI tool worth it? is for people who build with AI: developers, gamers, game makers and vibe coders trying to work out which subscription is actually worth the money. Prices and limits for these tools change every few weeks, and a list price alone doesn't tell you much. So we run the same build on each tool and report what it really cost.
The test bench
- Same brief for every tool. Each test uses a fixed, published brief (for example: "a small browser game with a title screen, score and restart" or "a to-do app with login"). Briefs are versioned so results can be compared over time.
- Real plans, our own money. We test on the paid plan the article is about, bought with our own account, not a press or free account.
- The real bill. For every run we log:
- Credits / usage, read from the tool's own usage dashboard
- Dollars: the plan share plus any overage or extra usage
- Minutes: wall-clock time from the first prompt to a working build
- Bugs: the number of issues we had to fix by hand or re-prompt for
- Sourced prices. Every price and plan detail links to the vendor's page, with the date we checked it.
- Placeholders, not guesses. Until a test has actually been run, the page shows [TEST RESULT PENDING]. We never estimate a result we haven't measured.
The verdict scale
| Verdict | Means |
|---|---|
| Worth it | For most of our audience, it pays for itself compared with the alternatives at the same price. |
| Worth it if… | Good value for a specific kind of user. We spell out the condition. |
| Skip | A cheaper or free option does the same job for most people. |
Updates
Each page shows when we last checked its prices. When a tool changes its pricing or limits, we re-check the page and, if needed, re-run the test and change the verdict. A change log is noted at the bottom of the article.
Who runs this
The site is owned and operated by Juliet Cardenas. Day-to-day research, drafting and price checks are done with the help of an AI assistant. A human checks every verdict before it's published. Disclosing that is part of being honest about how the site works.
Start reading
See our AI coding tool reviews and head-to-head comparisons, starting with Cursor vs Claude Code vs GitHub Copilot.
Money
Some tools pay us a commission when readers sign up; many don't. That never changes a verdict. Full details are in our affiliate disclosure.