How to choose a Pokémon card scanner

What does it do when it is not sure?

This is the question that matters, and it is the one no app puts on its store page. A scanner meets a card it cannot read — a glare across the holo, a sleeve, a worn corner — and it does one of three things. It guesses silently and shows you a price. It offers you a short list of the printings it thinks it might be. Or it tells you it could not read the card.

The first behaviour is the dangerous one, because you act on the number. A card named wrongly is a card priced wrongly, and you find out when you sell it for a fraction of what it was worth, or pay for something it is not. A wrong match shown without any warning leaves you to discover the mistake on your own.

Being cautious is not the same as being accurate, and it is worth saying so plainly: an app that asks you to confirm a printing has not proved it is better at reading cards. What it has done is limit the damage in the cases where no photograph can settle the question.

You can test this yourself, and the fair version of the test has two halves. Start with a card whose printing you already know and photograph it cleanly — a scanner that cannot get that right is not worth the second half. Then photograph the same card at an angle, in bad light, with the collector number in shadow. If it still returns one confident answer, check that answer. It may be right: a difficult photograph can still carry enough detail. What you are looking for is whether the app ever tells you when it is not sure.

Does it name the card, or the printing?

“Charizard” is not an answer. The same artwork exists across dozens of printings — different sets, reprints, holo and reverse holo, first edition and shadowless, secret rares numbered above the set total — and the gap between the cheapest and the dearest of them runs into hundreds of times, not percent.

So the useful output is the specific print: the set, the collector number, the variation. If the result hides those, you still do not know which printing the price belongs to, or how the app told it apart from the others.

What does a failed scan cost?

A free download does not tell you how repeated scanning is charged. Most scanners meter it — a number of scans a month, or a subscription. Read what happens when a scan does not work: if a photograph the app could not read still spends one of your scans, you pay even when you got no usable result, and a binder of awkward cards runs the meter fastest.

Where do the prices come from?

A price with no source behind it is a guess with a currency symbol in front of it. Ask which marketplace the figure comes from, how recent it is, and whether it is what cards sold for or what sellers are asking. Ask whether it changes with condition, because a near mint card and a played one are not the same asset.

What happens to your photograph?

Scanning a card means sending a photograph somewhere. Find out where it goes, how long it is kept, and whether an account is required to get an answer at all. An app that wants an email address before it will identify a card is collecting something other than cards.

Where GrailFox stands

The same five questions, answered about our own app. Where the answer is worse than we would like, it is written that way.

When it works, this is what you get. Photograph the card and GrailFox names the set, the collector number and the printing variation, then shows the value for that specific print in the condition you pick, with the marketplace the figure came from and when it was last updated printed beside it. It is an estimate of market value, not a price anyone has promised you.

When it is not sure, it says so. If the collector number cannot be established, GrailFox does not pick a printing for you — it offers the printings it considers possible and you choose, or it tells you the photograph did not read and suggests what to change. An exact match requires the matched card to carry the collector number that was read. That rule exists because our own testing once produced a confident wrong answer on an expensive card, and it is the single change that removed those answers from the test set.

A scan that fails costs nothing. A photograph that does not identify a card spends no free scan and no credit. That covers a scan the app could not read — it cannot cover a wrong answer the app believed, which is the case the collector-number rule is for and the reason the printing is worth checking against the card in your hand.

You get three scans before paying anything, and no sign-up: no email address, no password, nothing to create. Your collection is held against an anonymous identifier made on first launch.

Your photograph goes to OpenAI. This is the answer we like least, and leaving it vague would make the question above dishonest. GrailFox sends the photograph to OpenAI, which reads the card's name, number and printing clues from it; the request carries no identifier of yours. We keep no copy on our servers. Under our current settings OpenAI may keep photographs and use them to improve its models, and we have no special deletion arrangement with it. If you add the card to your collection, a copy of that photograph stays on your phone so the collection shows the card you own. The full detail is in the Privacy Policy.

What it does not cover. English Pokémon cards on iPhone. Not Japanese printings, not other trading card games, not sports cards, and there is no Android version. It does not predict what a grading company would give a card — a phone camera cannot see sub-surface creasing or measure a corner, and a number that looks like a grade would be false confidence on exactly the cards you would send away. It does not sell anything for you.

Try it on a card whose printing you already know, and check the set and collector number against what you are holding before you trust the price.