How We Test AI Tools

Every recommendation on AiSuperSmart follows the same rules. This page explains exactly how we evaluate AI tools, what our labels mean, and what we do when we get something wrong.

Two kinds of articles, clearly labeled

We publish two types of tool coverage, and we tell you which one you are reading:

  • Hands-on reviews. We used the tool ourselves. These articles include our own screenshots, the exact tasks we ran, how long we used the tool, and what we paid. If a review says “we tested,” there is evidence in the article to back it.
  • Research-based comparisons. For fast-moving roundups, we compare official specifications, published pricing pages, free-tier limits, and independent benchmark data such as the Artificial Analysis arena rankings. These articles say so in a “How we picked” section, and we never imply first-hand use we did not do.

What we verify before publishing

  • Pricing and free-tier limits come from the vendor’s official pricing page at the time of writing, and we state the date. When sources conflict, we say so instead of guessing.
  • Statistics are cited inline with a link to the original source. We use primary sources, established research firms, and major publications. If we cannot verify a number, it does not go in the article.
  • Vendor benchmark claims are always labeled as self-reported.
  • Every external link is checked at publish time.

How tools earn a spot in a ranked list

We weigh five things: what the tool actually does compared to its marketing, real pricing including hidden limits, the generosity and honesty of the free tier, independent quality signals (benchmarks, arenas, credible user reports), and who the tool is genuinely best for. Every ranked list names a specific “best for” use case per tool, and includes at least one honest weakness per tool. A tool no one should use does not get a spot just to pad a list.

Freshness and updates

AI tools change weekly. We update articles when facts materially change — a price change, a shutdown, a major model release — and we only update the “last updated” date when the content substantively changed. High-churn pages (pricing trackers, usage-limit guides) carry a visible changelog.

Corrections

If we got a fact wrong, we fix the article and note the correction. Spotted an error? Use our contact page and we will review it within a few days.

Affiliate disclosure

Some links on this site are affiliate links, which means we may earn a commission if you sign up — at no extra cost to you. Affiliate status never changes a tool’s ranking, rating, or the weaknesses we report. We recommend free and non-affiliate tools whenever they are genuinely the better pick.

Scroll to Top