Tony Wang

Posted on Jul 27 • Originally published at crawlora.net

How Many AI Apps Are on the App Store? 1,490 Analyzed (2026)

#ai #datascience

Key takeaways

The gold rush is real: of ~1,490 discoverable 'AI' apps on the App Store, 84% launched in 2023 or later — more in 2023 alone (488) than in every prior year combined (236).
But the takeover skipped the top chart. Only 6–8% of the Top 100 free apps are AI — the same handful (ChatGPT, Claude, Gemini, Meta AI) on both Apple and Google. The flood is in the catalog, not the rankings.
It's a free-install, subscription business: AI is 0% of the top paid chart, yet ChatGPT is the #1 top-grossing app on the iPhone.
Five apps, half the ratings. The distribution is brutally unequal (Gini 0.95): ChatGPT alone holds 22% of all AI-app ratings and the top 5 hold 45%. The median AI app has just 171 ratings; two-thirds have under 1,000.
Built by an army of one-app developers: ~1,300 developers for ~1,490 apps, 90% shipping a single app, a quarter carrying template 'AI…/Chat…' names. 30 of the top 50 results for 'chatgpt' aren't even made by OpenAI.
Honest caveat: most of these apps aren't bad — they're ignored. The long tail is unrated, not one-star, and we make no quality claim from store ratings.

Everyone agrees AI ate the App Store. The interesting question is where it took a bite — and the answer is not the one the headlines imply. So instead of vibes, we pulled the listings: about 1,490 distinct "AI" apps from Apple's App Store, the full top charts across nineteen categories, Google Play's install counts, and a sample of reviews — all in June 2026.

How many AI apps are on the App Store? About 1,490 are unambiguously AI as of June 2026 — out of roughly 2.42 million apps total — and that is just the discoverable head, so the real figure is higher. Don't confuse it with the web AI-tool directories that claim anywhere from 1,200 to 90,000+: those count websites and browser tools, not installable App Store apps, which is a different and much fuzzier question. For scale, AI apps booked about $18.5 billion in revenue in 2025 with over 1.1 billion users — this study is about which of those ~1,490 actually broke through.

A method note, with the usual irony: we read the stores through a structured web-data API. Both Apple and Google rate-limit and fingerprint plain scrapers, and Google Play hides install counts behind rendered pages — which is exactly why a maintained, structured API beats hand-scraping. (The short version is in the box below; full caveats are at the end.)

Methodology at a glance

Storefront & window: US App Store + Google Play, a single June 2026 snapshot.

Corpus: ~1,490 distinct "AI" apps, paginated from about a dozen AI seed searches — the discoverable head of the market, not a census, so every concentration figure is a lower bound.

What counts as "AI": a strong signal only — an explicit "AI"/"GPT"/assistant marker or a named AI brand — not the bare word "chat" (which over-counts messaging and dating apps).

Rating counts are a proxy for scale, not exact installs (Apple doesn't publish installs).

Google Play install counts are Google's own bucketed figures, inflated by pre-installed apps like Gemini and Copilot.

Review quotes are a hand-read sample for illustration, not a sentiment census.

Full caveats are at the end. Every figure came back as structured JSON — see a sample App Store or Google Play response.

What this means

The story isn't "AI is winning the App Store" — it's that AI flooded the catalog while the charts barely moved. Supply exploded; attention didn't. What that implies depends on who's reading:

App builders: a generic chatbot wrapper now joins ~1,490 look-alikes where the median app has just 171 ratings — a crowded, low-traction bet. The opening is category-specific workflows in the verticals AI has barely touched (Finance, Shopping, Navigation and Social were all 0–2% of their top charts).
Investors & market researchers: read catalog counts as noise and traction as signal. The boom is supply, not demand — five apps hold ~45% of all AI-app ratings and ChatGPT alone holds 22%. Judge AI apps on ratings, installs and grossing rank, not on how many shipped.
ASO teams: the top of the chart is locked by four incumbents, but the long tail of 'chatgpt'-intent searches still moves — 30 of the top 50 results for that query aren't OpenAI. Track rank, ratings and review sentiment continuously, because the wrapper tier turns over fast.
Data & API buyers: the durable signal is rankings, install counts and reviews pulled as structured JSON and re-pulled on a schedule — a one-time snapshot like this one decays as the charts shift.

The gold rush is real

Start with supply. Tag every app whose name or maker signals AI, bucket by original release date, and the category simply materializes after ChatGPT:

84% of the AI apps we found launched in 2023 or later — more in 2023 alone than in the App Store's entire prior history of AI. That tracks the wider market: Appfigures put 2026 app releases up 60% year-over-year across both stores (and +80% on iOS alone), and Sensor Tower clocked AI-app downloads up 148% in 2025. AI didn't just grow; it became the default thing to ship.

But it skipped the top of the chart

Here's the twist. All that supply barely dented the rankings. On any given day, only a handful of the Top 100 free apps are AI-native — and it's the same handful on both stores: ChatGPT, Claude, Gemini, and Meta AI. The rest of the chart is the incumbents you'd have seen in 2019: streaming, banking, shopping, social.

What's distinctive is how AI monetizes. It's almost never a paid download — it's a free install wrapped around a subscription:

And the takeover is wildly uneven by category. AI saturated creative tooling and assistants, and barely touched anything transactional:

Four apps ate the AI

Now the part that reframes everything. Across both stores, just four AI-native apps crack the Top 100 free chart — ChatGPT, Claude, Google Gemini, and Meta AI — and the ratings show how empty the space behind them is. Rank all 1,490 apps by rating count and the curve is almost vertical: ChatGPT alone holds 22% of every AI-app rating we counted; the five most-rated — ChatGPT, Canva, Google Gemini, Gauth, and Microsoft 365 Copilot — hold 45%; the top hundred hold 91%. (The two lists barely overlap — Claude and Meta AI top today's charts but trail on lifetime ratings, where big installed bases like Canva, Gauth and Copilot pile up; a rating count rewards reach and age as much as hype.)

The flip side of that curve is a vast, silent tail. The median AI app has just 171 ratings. Two-thirds have fewer than a thousand. Seven have more than a million. ChatGPT, with nearly eight million, has roughly forty-six thousand times the median app:

That's the whole shape of the gold rush: a few prospectors struck a seam, and everyone else is panning an empty creek.

An AI app for every feeling

What are the other 1,480 apps? Mostly the same few ideas, copied. Cluster the corpus by what each app does and it's a narrow menu — assistants, photo generators, and an enormous pile of AI companions:

An app farm, not an ecosystem

The 1,490 apps come from about 1,300 developers — and 90% of them ship exactly one app. Nearly a quarter of all titles match a template name (they start with "AI " or are some permutation of "Chat / Bot / Assistant / GPT"). This isn't a thousand teams solving a thousand problems; it's a thousand wrappers around the same few APIs, racing to rank for the same keywords.

You can see the keyword race most clearly around one word: ChatGPT. Apple polices literal name-spoofing — the fake "ChatGPT 5.0 Official" apps that surface in search autocomplete don't survive into the ranked results. So the clones adapted: instead of copying the name, they copy the intent. Of the top 50 results for the query "chatgpt," 30 are not made by OpenAI — a combined 3.7 million ratings, all free, fronted by anonymous LLCs and overseas studios wrapping the OpenAI API behind a paywall (the busiest of them is the very same one-developer "app farm" that tops our corpus).

And the reviews are where the wrapper model shows its seams. These are recent reviews — the apps' lifetime ratings stay high — but the same complaint recurs across the clone tier: a surprise subscription you didn't mean to buy.

"I accidentally paid for this app instead of the ACTUAL Chat GTP that I can actually CHAT with…" — one-star review of a ChatGPT-wrapper app

"I was charged around $200 AUD for this app despite barely using it… I didn't even recognise the name when the charges appeared and had to Google it to work out what it was." — one-star review of the same app

"They initially said it is free of charge and the credit card is used only for age verification; once the card details are given, immediately they charge USD 39.99." — one-star review of a 'ChatGPT' clone

To be clear, this is the pattern at the wrapper tier, not a verdict on AI apps generally — plenty of users love the real thing ("ChatGPT is truly the best — I cannot imagine how I lived life without it"). But it's the predictable economics of a market where a thousand near-identical apps compete for installs and monetize on the way in.

Android tells a slightly different story

One more cut, because Google Play exposes something Apple won't: real install counts. Run the same analysis there and the concentration is just as steep — but the king is different. On Android, Gemini, not ChatGPT, leads, and four apps clear a billion installs:

It's a neat illustration of home-field advantage — Google ships Gemini with Android, the way Apple would if it had a comparable app — and a reminder that "most popular AI app" depends entirely on which store you measure.

How we did this (and the caveats)

We pulled App Store and Google Play listings, charts, and reviews through a structured API in June 2026, on the US storefront. The AI corpus is ~1,490 unique apps gathered by paginating searches across a dozen AI seed terms; the chart and category figures come from the live top-100 lists. Caveats worth stating plainly: "AI app" is a tagging judgment — we required a strong signal (an explicit "AI"/"GPT"/assistant marker or a named AI brand) rather than the bare word "chat," which over-counts messaging and dating apps. The corpus is the discoverable head, not a census — store search has a relevance floor, so the true long tail is even longer and the real inequality even higher than the Gini of 0.95 we measured. Release dates are the original bundle's first release, so apps that added AI later (Canva, Picsart, Snapchat) read older than their AI debut; the pure-play 2023+ chatbots are the clean "AI-native" cohort. Rating count is a proxy for scale, not exact installs (which Apple doesn't expose and Google reports in buckets). And we make no claim that low-rating apps are bad — most simply have no traction, which is a different thing from a bad review.

If you're wondering how we read charts and reviews that block plain scrapers: that's the day job. Crawlora is a web-data API for AI agents and pipelines that returns normalized JSON for the App Store, Google Play, search engines, marketplaces, and social platforms — handling proxies, rendering, and anti-bot — and bills pay-on-success. The same approach powers our Product Hunt trends study, our streaming fragmentation study, our podcast gold rush study, and our anti-bot adoption index — and the full app catalog behind this study, 4.1M apps across the App Store and Google Play, is browsable in our mobile app dataset — with the daily App Store + Google Play charts and app-review sentiment as their own queryable datasets.

Sources

Frequently asked questions

How many AI apps are on the App Store?

We counted about 1,490 unambiguously-AI apps on the US App Store in June 2026 — out of roughly 2.42 million apps in total — and that is only the discoverable head, so the true figure is higher. (Don't confuse this with web 'AI tool' directories that cite anywhere from 1,200 to 90,000+; those count websites and browser tools, not App Store apps.) It's a long tail: 84% launched in 2023 or later, the median app has just 171 ratings, and two-thirds have under 1,000.

What is ChatGPT's market share among AI apps?

Among App Store AI apps, ChatGPT holds about 22% of all ratings and is the #1 top-grossing iPhone app. It is not unrivalled: Gemini leads on Android installs (helped by OS bundling), and across the wider AI-assistant market ChatGPT's usage share recently slipped below 50% as Gemini, DeepSeek, Claude, Grok and Perplexity grew. Either way the concentration is extreme — the top five apps hold about 45% of all AI-app ratings.

How many AI apps are there like ChatGPT?

Many, and most are wrappers. Of the top 50 App Store results for 'chatgpt,' 30 are not made by OpenAI (a combined 3.7 million ratings), and about a quarter of all AI apps carry template 'AI…/Chat…' names. But attention is concentrated: ChatGPT alone holds about 22% of all AI-app ratings, and the next assistants — Gemini, Claude, Perplexity, Grok, DeepSeek — trail far behind on the App Store.

What is the most popular AI app?

ChatGPT, by a wide margin. On iOS it has nearly 8 million ratings — about 22% of all AI-app ratings we measured — and it is the #1 top-grossing iPhone app. On Android, Google Gemini leads on installs (~1.9 billion), helped by being bundled with the OS.

Is the App Store full of AI apps now?

In supply, yes; on the charts, no. Only about 6–8% of the Top 100 free apps are AI, and it's the same handful (ChatGPT, Claude, Gemini, Meta AI) on both Apple and Google. The flood is in what gets listed, not what ranks.

Which app categories did AI take over?

Mostly creative tooling and assistants. AI is about 28% of the Photo & Video top chart and 27% of Productivity, with smaller footholds in Reference (translators) and Music. It is essentially absent — 0% — from Weather, Shopping, Navigation, Games and Food & Drink.

Are most AI apps any good?

Most simply get no traction rather than bad reviews — the median has only 171 ratings. About a quarter are template 'AI…/Chat…' wrappers, 90% come from single-app developers, and 30 of the top 50 results for 'chatgpt' aren't even made by OpenAI. We make no quality claim from store ratings, which stay high for the apps people actually use.

How was this measured?

By pulling App Store and Google Play listings, top charts, and reviews through Crawlora's structured web-data API (US storefront, June 2026) and aggregating them. The stores rate-limit plain scrapers and Google Play hides install counts, so a structured API is how you read them at scale. Figures are a point-in-time snapshot and a lower bound on the true market.

Originally published on crawlora.net. Crawlora is a structured web-data, search, and anti-bot API — dozens of platforms as normalized JSON, plus a hosted MCP server, with a free tier (no card).

DEV Community