The AIFE Score is simple enough to repeat to a friend — and we publish the whole thing. No sponsored ranks, no hidden weights, no filled-in blanks.
Two lenses, each anchored to real data. The percentages are the actual weights in the formula — published from the engine code, not approximated.
Verifiable crowd behavior — every trusted ranking (market cap, players, votes) defaults to it.
Source: Monthly website visits (Similarweb + Tranco), domain-attribution corrected
Live — monthly visits, log-scaled against the whole board
Separates the merely-big from the actually-good; lifts Claude above higher-traffic, lower-graded rivals.
Source: Graded review across 5 dimensions (features, performance, ease of use, use cases, support)
Live for graded tools; grading pipeline expanding to the top 300
These are Claude’s live numbers as of 2026-07-15 — check them against the table yourself.
Clauderanks above tools with more traffic (Grok, DeepSeek) because its quality grade lifts it — that’s the 30% doing its job.
A tool missing either lens shows “Not scored yet”— never an imputed or renormalized number. We’d rather show fewer scores than fake one.
Real traffic data and a real quality grade exist. The formula above runs exactly as published — nothing else touches the number.
Either lens is missing. No blank is filled with a guess, an average, or a re-shared weight. The tool still ranks normally by popularity and price in the table while we close the gap.
Quality is graded across 5 dimensions. A tool needs at least 3 of the 5 to count — an isolated single grade is noise, not signal.
Monthly visits from Similarweb and Tranco, log-scaled against the whole board. We correct domain attribution by hand: a sub-product does notinherit its parent domain’s traffic (a tool hosted on a giant’s domain doesn’t get the giant’s numbers). Where a value is an estimate rather than a measurement, it’s marked with ~ in the table.
An AI-assisted grading of the tool’s own product information across the 5 dimensions above (methodology id ai-assisted-v1). Said plainly: this is not member reviews yet. Real user reviews replace this lens as they accumulate — see the roadmap below.
When we fitted the formula against real-world rankings, price carried roughly zero weight — so it isn’t in the Score. It has its own column in the table; folding it into the number as well would double-count it.
A “Coming” lens affects nothing until its pipeline ships. And a standing rule: votes and saves will never move a rank.
Coming — real 7-day traffic movement from daily Tranco snapshots (pipeline accruing)
Coming — member reviews & saves once a real base exists; votes never move a rank
Same formula, applied uniformly, to every tool. We don’t sell placements, and no payment of any kind can move a tool up the page. The weights on this page are the weights in the code — if they change, the changelog below says so, dated.
AIFE Score v3 · 2026-07-15 — Reverse-engineered two-lens rebuild: 70% Popularity + 30% Quality, strict gate replaces renormalization, price removed from the score, domain-attribution pins written to visits data.
AIFE Score v2 · 2026-07-15 — Rebuilt for honesty after auditing how Artificial Analysis, LMArena, G2 and a16z score: no blank is filled with a neutral guess, the score is anchored in real adoption, and every tool ships a confidence badge. We publish the mix and the math — the only AI-tools directory that shows you both.