Which identification apps are actually free?

Why the free ones are free

The pattern in the table is not a coincidence and it is worth understanding, because it predicts things the table does not show.

Merlin and eBird come from the Cornell Lab of Ornithology, a membership-funded university research programme. The Audubon guide comes from a conservation non-profit. PlantNet comes from a consortium of French public research institutes. Seek and iNaturalist come from a non-profit joint initiative. For all of them, identification is the institutional mission and the app is outreach — a paywall would work against the reason the organisation exists.

That funding model also explains their quality, which is the part people find surprising. Merlin’s live sound identification is not matched by any paid app, because it rests on the Macaulay Library of expert-verified recordings and on eBird’s record of what is actually present near you. A commercial publisher can build a good classifier; it cannot build a century of ornithology.

The commercial apps are not thereby bad. They are better at cultivars, at care guidance, at answering every time, and at interface polish — all real things. But if your question is “which is best”, the honest answer in birds, and arguably in wild plants, is a free one.

How to read an in-app purchase list

Apple publishes the name and price of every in-app purchase on an app’s product page, and it is the most useful thing on that page. It does not publish the billing period, which is where the confusion comes from.

So a tier called “Premium” at a low price can be weekly, monthly or annual and the listing looks identical either way. Read the list as a shape rather than as prices: several tiers sharing a name is one product billed over different terms, the lowest number is almost always the shortest term, and it is the one the app presents first because it looks cheapest.

Do the arithmetic before you tap. A weekly subscription at a few dollars is several hundred a year — routinely five times the annual tier of the same product. If a free trial converts to weekly billing, that is the tier to avoid, and the annual one is usually two taps away in the same sheet.

Two defences worth knowing. Look at the in-app purchase list before installing, when you are not three screens into an onboarding flow with a countdown on it. And if you were charged for something you did not intend to buy, request a refund at reportaproblem.apple.com — Apple grants those more often than people expect, particularly for a first accidental subscription.

Cancelling, which deleting does not do

Subscriptions live with your Apple Account, not with the app, so removing the app leaves the billing running. This is the single most common mistake in this category.

iPhone or iPad: Settings, tap your name at the top, then Subscriptions. Every active subscription across every app is in that one list, and it is authoritative regardless of what an app’s own settings screen says. Mac: App Store, your name at the bottom of the sidebar, View Information, then Manage. Android: Play Store, profile icon, Payments and subscriptions.

Cancelling keeps the subscription working until the end of the period you already paid for, so there is no reason to delay it.

What no app in this table does

Every free identifier here — Seek, iNaturalist, PlantNet, Merlin, eBird, Audubon — is built on a taxonomy of living things. A rock has no taxonomic position. Neither does a fossil, a coin, an arrowhead or a mark on the base of a jug.

That is not a gap more training data would close; it is what those apps are, and it is the right scope for them. But it means the free, institution-backed option that exists so well for plants, birds and insects has no equivalent for the other half of what people photograph. There is no Cornell Lab of geology shipping a free scanner.

Some of that half is not a camera problem at all. A mineral is identified by streak, hardness, cleavage and heft; a track by counting toes and looking for claws; a mushroom by a spore print left on paper overnight. Those are tests rather than photographs, which is why the free tools on this site ask you to do them — determination keys and field calculators, in a browser, with no account and nothing to buy.

What is not in this table, and why

Three categories were left out deliberately, and saying so is more useful than a longer table.

Plant-care apps with identification bolted on. Planta, Blossom, Plant Parent and several others are genuinely good products for keeping a houseplant alive, and identification is the front door rather than the room. They appear in individual comparisons where relevant, but ranking them as identifiers misrepresents what they are.

Apps with a famous name and a different publisher. The App Store listing carrying a name you remember is not always the product you remember — the developer line under the app name is the most informative field on the page, and it takes ten seconds to read. Our LeafSnap page works through the clearest example.

Anything we could not price from Apple’s own data. If a figure could not be read from the Lookup API or the product page, the app is not in the table, because the alternative is an estimate — and an estimate presented next to verified figures is worse than an omission.

How these pages are kept honest

Three rules, enforced by the build rather than by intention.

No number is typed. Every price, tier, rating, review count, version, size and update date on every page in this section is read from Apple’s Lookup API and product pages by a script, stored as a dated snapshot, and substituted at render. The build fails if a price appears in the editorial text, because a price a human typed is a price that goes stale silently — and on a comparison page a stale price is worse than no page at all.

The snapshot date is printed everywhere. Prices in this category move, and a figure without a date is a claim without evidence.

Where a competitor is better, the page says so. Orvik is our app. It appears in some of these comparisons and it loses several of them, which is the only way a comparison page written by an interested party is worth reading.