Skip to content
AI Tier ListField guide
01 · METHODOLOGY

How we rank AI tools

AI Tier List grades 233 AI tools in 20 categories into five tiers, S to D. Four inputs decide it — capability on the category's core job, pricing relative to what you get, maintenance activity, and adoption signal — and the tier is an editorial judgement on them, not a computed score. A pipeline re-evaluates every week, tiers move down as well as up, and no tier is for sale.

This page answers the fair objection that a tier is just an opinion. What we look at, how often we look again, which data we use, and where the money comes from — all of it is written down below.

LAST UPDATED ·

1. What the tiers mean

Every tool carries one of five grades: S, A, B, C, or D.

S
Best in classStart here in this category
A
ExcellentThe alternative when S doesn't fit
B
GoodWorks for specific needs
C
AverageBetter options exist
D
BasicOnly when nothing else is available

One thing matters more than the labels themselves: a tier is relative to the tool’s own category, never across categories. An S-tier music generator is not "better than" a B-tier coding assistant — it is the first thing to try if the job is generating music. The two are not doing the same job, so there is no single ladder that holds both.

2. How a tool gets rated

We look at four things when placing a tool:

  • Capability against the category’s core job — how well it actually does the thing the category exists for. The length of the feature list is not the measure; the output is.
  • Pricing relative to what you get — not the absolute number, but what the number buys: how usable the free tier really is, where the paywall sits, and how predictable usage-based billing turns out to be.
  • How actively it is maintained — an abandoned tool falls behind fast in this field. We check release notes, how quickly it adopts newer models, and whether the docs still describe the product that ships.
  • Adoption signal — whether people actually use it. We track search interest over time and, for models, real usage rankings. Popularity is not quality, but a tool nobody uses is usually telling you something.

To be plain about it: the final tier is a judgement informed by these inputs, not a computed score. There is no weighted formula summing these into a number. We are not going to invent one — a score to two decimal places looks precise, but the weights behind it were still chosen by a person, and that choice is the part directories never publish. We would rather publish the inputs and admit a human made the call.

3. How often it gets re-checked

An automated pipeline runs weekly. It re-evaluates the tools already listed, collects search interest per tool per week, and takes a weekly snapshot of OpenRouter model usage rankings. Every tool page shows the date it was last re-ranked, so you can see how old the tier in front of you is.

Tiers move in both directions. A list where nothing is ever downgraded is not a ranking. Prices go up, free tiers shrink, competitors ship past, development stalls — and the grade drops accordingly.

4. Data sources

  • Google Trends — weekly search interest per tool. This is what the trend badges and charts on tool pages are built from.
  • OpenRouter usage rankings — weekly snapshots ranked by actual API spend, not by votes or a survey. This is what powers the LLM rankings.
  • Vendor documentation and pricing pages — features, limits, and prices are recorded from the vendor’s own published pages, not from secondhand summaries.
  • Hands-on and editorial review — the part the other three cannot see. Some things you only learn by using the tool.

We also use AI, and say so. It assists the scanning for new tools, the collection of changes, and the drafting of copy. High-confidence changes are applied automatically, without a person in the loop. There is no reason to be vague about this, so: when the pipeline judges the evidence strong enough — a price rise, a shrunken free tier — the change goes live immediately. Changes below that confidence bar are not applied; they are held for review instead. Every change is logged with what moved, when, and on what evidence, and shows up in the recent-changes section of the tool page. If you spot one that is wrong, tell us below — in this setup that is the fastest correction path there is.

5. How this site makes money

This site runs on display ads and affiliate commissions. Some tool links are affiliate links, and we get paid when you sign up through one. You pay the same either way. That is the whole business model — we would rather state it up front than bury it.

The obvious problem with that model is the conflict of interest it creates: there is real pressure to grade the tools that pay us more generously. So we hold to three rules.

  • Tiers are not for sale. No paid placements, no sponsored slots in any ranking.
  • Whether a tool runs an affiliate program is not one of the criteria. The criteria above are all of them. In practice many of our top-tier tools run no affiliate program at all and pay us nothing.
  • Every commission-earning link is labelled as such right next to the link. You should know which links pay us before you click, not after.

We do not accept offers to move a tier, and receiving one would not move it. If you ever spot a grade that looks like it followed the money, tell us below.

6. If you disagree with a tier

If a grade looks wrong to you, say so. A tier rated against published criteria should be arguable against those same criteria. Send the tool name and why you think the grade is wrong, and it goes into the next re-evaluation.

seanjr28475@gmail.com

For a broader description of the site, see the about page. To see the criteria applied, open any category tier list or a head-to-head comparison.

Questions about the rating system

Is a tier a computed score?

No. A tier is an editorial judgement on four inputs: capability on the category's core job, pricing relative to what you get, maintenance activity, and adoption signal. There is no weighted formula. A score to two decimals looks precise but only hides that a person chose the weights, so we publish the inputs instead.

How often do tiers change?

An automated pipeline runs weekly, re-evaluating listed tools and collecting per-tool search interest and OpenRouter model-usage snapshots. Every tool page shows its last re-rank date. Tiers move both ways: a price rise, a shrunken free tier or stalled development pushes a tool down.

Can a vendor pay for a higher tier?

No. Tiers are not for sale; there are no paid placements or sponsored slots. Whether a tool runs an affiliate program is not a criterion, and commission-earning links are labelled next to the link. Many of our top-tier tools have no affiliate program and pay us nothing.

Is an S-tier music tool better than an A-tier coding tool?

That comparison does not hold. A tier is relative to the tool's own category. An S-tier music generator is the first thing to try when the job is generating music, not a claim that it beats a coding assistant doing a different job.

What if I disagree with a tier?

Send the tool name and why you think the grade is wrong to seanjr28475@gmail.com. A tier rated against published criteria should be arguable against those same criteria, and every dispute goes into the next weekly re-evaluation.