Claims

These are marketing claims. No published test, no count, no method: treat the number as fiction.

FALSE

Their own numbers say otherwise

UNPROVEN

No test, no count, no method

MISLEADING

The number isn't what they say it is

BACKED

Counts and misses published

Vendors: put your engine on the board.

Every engine here ran the same 1,000 photos under the same rule. Send us API access and yours runs the same test. Results are published.

Submit your API

Accuracy claims

They say	Verdict	Why	Who
“Ludex is the fastest, most accurate scanner for sports and trading cards”	UNPROVEN	No test. No numbers. No comparison. It's a slogan.	Ludexludex.com · 2026-09-29
“Scan+ detects about 95% of cards”	UNPROVEN	A support reply. No test set, no count, no method.	CollXdeveloper reply on Google Play · 2026-09-29
“98.7% match rate with professional human grading”	UNPROVEN	No sample, no method, no data. Their own blog says version 1 was shut down for inconsistent results.	Digital Grading Companyvendor blog · 2026-10-03
“92.8% accuracy, ±0.22 grade variance”	UNPROVEN	Press release only. Their own materials can't agree on 47 inspection points or 16 zones.	CardGrade.iolaunch press release · 2026-10-03
“95% alignment with PSA grades”	UNPROVEN	No count, no sample, no method.	CardMintAIcardmintai.com · 2026-10-04
“Independent testing puts Ludex at ~85% and CollX at ~65%”	UNPROVEN	No such test is published. The 10,000-card study it cites can't be found anywhere.	PreGradeCardspregradecards.com · 2026-10-21
“Roughly 99% accuracy on the first scan”	FALSE	Their own test: 96%, on 20 cards. They market 99%.	Eyevoproduct pages · 2026-10-21
“More than 700 million cards scanned with over 98% accuracy”	MISLEADING	Those percentages are model confidence, not accuracy. No measured rate on any published set.	TCGplayerseller pages · 2026-10-21
“95.5% recognized on 1,000 Pokémon images: 955 right, 25 held, 20 wrong”	BACKED	955 right, 20 wrong, 25 held. Every answer published.	BlazeLensblazelens.ai, published 2026-10-15 · 2026-10-15