Cal AI is the app that made “point your camera at the plate” mainstream, and it earned that on the basics. It's also had the strangest year of any app in this category, and most of the “alternatives” articles written since respond to it in exactly the same way: here is another app that points a camera at a plate.

Short answer

Cal AI is now owned by MyFitnessPal (deal announced March 2026, app still standalone), was temporarily removed by Apple in April 2026 over deceptive paywall and subscription design, and is back on the store at 4.8 stars. Its photo accuracy is roughly the same as every other photo app — fine on separable foods, weak on portions and sauces. If you want the same workflow with cleaner billing: Snap n Eat (Australian) or MyFitnessPal's own Meal Scan. If the problem is what a photo can't see, the alternative is to say the meal instead — which is what Rekkon does.

This one doesn't, for a reason I'll get to. But first, what actually happened.

#Cal AI in 2026, briefly

What changed, and when (as reported)
WhenWhat
Dec 2025Acquired by MyFitnessPal; deal closes (per TechCrunch)
2 Mar 2026Acquisition announced. Cal AI stays a separate app; team retained; food data integrated with MyFitnessPal's database
Apr 2026Apple temporarily removes Cal AI, citing deceptive billing design and manipulative subscription flows (per TechCrunch)
Jun 2026Reinstated after changes; live on the App Store
Aug 2026Around 4.8 stars across roughly 349,000 ratings

So: a popular, well-rated app now owned by the biggest name in the category, with a documented billing problem that Apple thought serious enough to act on. Reviewers still flag that pricing sits behind the onboarding quiz, the listing carries several different “one month” prices, and the short trial converts automatically. If you're on it and happy, stay — just cancel through the App Store, not by deleting the app.

#The thing every alternatives list gets wrong

Search “Cal AI alternatives” and you'll get a dozen articles, nearly all written by other photo apps, each concluding that the alternative to Cal AI is them. That's not dishonest exactly, but it skips the question that matters: why are you looking?

If it's the billing, fine — pick a photo app with cleaner pricing and carry on. But most people who drift away from photo logging do it for a different reason. The number was wrong, repeatedly, on the meals that mattered. And that isn't a Cal AI problem. It's a photograph problem.

#What a photo can't see

We've written about this at length in the accuracy piece, so the short version. Photo food recognition is decent at identifying what is on a plate — research puts it around 68 to 86 percent on common foods. It is poor at how much: portion-size estimation drops as low as 39 percent, and it's worst exactly where restaurant plates and home cooking live — depth of bowl, size of plate, food stacked on food.

And there's a whole category it can't see at all: the oil the chicken was cooked in, the butter under the vegetables, the dressing already mixed through the salad, the second glass of wine that was poured after the photo. These aren't edge cases. They're the calories that make the number wrong.

Every photo app inherits the same ceiling. Switching from one camera to another changes the interface, not the physics.

#The alternatives, sorted by why you're leaving

Leaving because of the billing or the ownership? The same workflow with fewer surprises:

Leaving because you need the numbers to be right? Cronometer, with verified lab-analysed data and micronutrients. It has photo logging on Gold, but its strength is what happens after the entry. Read our Cronometer piece for when that's the right call.

Leaving because the number kept being wrong on real meals? Then the alternative isn't a camera. It's telling the app the things the camera can't see.

#Describing beats photographing, for a specific reason

When you say “chicken stir fry, cooked in oil, big bowl of rice, and a schooner”, you've supplied four things no photo contains: the cooking fat, the portion, the drink, and your own sense of scale. Voice logging is still self-report and still underestimates — by roughly the same 10 to 20 percent every method does — but its error is on the things you forget to say, not on the things you physically cannot show.

That's the case for Rekkon, which we build, and the limits are worth stating as plainly: iPhone only, paid after a seven-day trial, calories and macros but no micronutrients, estimates that are directional rather than lab-grade. What you get is a five-second log with no camera, no card to start, no account, and coaching that reads your actual day back — “you've had 1,650, dinner should be light, here's how.”

Say what the camera can't see.

“Chicken stir fry, cooked in oil, big bowl of rice.” Rekkon estimates it and coaches from there. Seven days free, no card required.

Try Rekkon free

#Quick decision

About this article. Written and reviewed by , founder of Rekkon — a competing app, named wherever it appears. Facts about other apps come from their own store listings and websites, and from cited reporting, as of September 2026; features and prices change, so check the current listing. General nutrition information, not medical advice.

#Common questions

Who owns Cal AI now?

MyFitnessPal. According to TechCrunch's reporting, the deal closed in December 2025 and was announced on 2 March 2026; Cal AI continues as a separate app, its small team was retained, and its food data has been integrated with MyFitnessPal's database. What that means for future pricing or a merge hasn't been disclosed.

Why was Cal AI removed from the App Store?

In April 2026 Apple temporarily removed Cal AI, citing — per TechCrunch — deceptive billing design and manipulative subscription flows. The app was reinstated after changes. Reviewers still note that pricing sits behind an onboarding quiz, its listing shows several different monthly prices, and the short trial converts automatically, so keep your own record if you subscribe and cancel through the App Store rather than by deleting the app.

Is Cal AI accurate?

About as accurate as any photo tracker: good on clearly separated foods, unreliable on mixed dishes, sauces, hidden oils and portion size. Research puts photo food identification around 68–86% and portion estimation as low as 39%. Reviewers report roughly 80% on a first photo, improving after you correct it. We wrote a full breakdown in our Cal AI accuracy article.

What is the best Cal AI alternative?

For the same photo workflow with clearer billing: Snap n Eat if you're in Australia, MyFitnessPal Meal Scan (which now owns Cal AI's technology), Lose It's Snap It or MacroFactor's photo AI. For micronutrients, Cronometer. For coaching and no camera at all — describe the meal and be coached on your day — Rekkon, which we build. It's iPhone only and paid after a trial.

Why not just use a better photo app?

Because the weakness isn't the app, it's the photograph. A camera cannot see the oil the chicken was cooked in, how deep the bowl is, or whether that's a small or large plate. Every photo tracker inherits that ceiling. Describing the meal lets you say the thing the camera can't see — “cooked in oil” costs three words.