SnapCalorie review

SnapCalorie access and features at a glance

The company behind SnapCalorie

Perception Labs, Inc. appears as seller and developer in the App Store listing. SnapCalorie's privacy notice names the company as doing business as SnapCalorie. Its FAQ says several team members are former Google AI researchers. The App Store description says its founders co-founded Google Lens and Cloud Vision API.

Apple's release record dates the first App Store release to November 4, 2021. The App Store record dates version 141.0.4 to October 8, 2026. Google Play also lists version 141.0.4, with an update date of October 7, 2026.

From a photographed plate to a nutrition entry

The SnapCalorie home page starts with a meal photo taken in the app. Its App Store description says AI identifies the foods and measures portions before returning calories, macros and more than 100 nutrients. The description attributes measurement to computer vision and depth sensors, with a separate LiDAR claim for iPhone Pro. The Google Play description keeps the depth sensor claim but omits that LiDAR paragraph.

The FAQ describes a phone depth scan of food volume. Hidden oils and sugars are guesses based on typical amounts for the dish and visible clues, according to the same FAQ. Depth sensor and LiDAR measurement are SnapCalorie's descriptions of its method, not measurements we checked.

For details a photo may leave out, the FAQ allows typed hints such as a menu description or known ingredients. The App Store description says cooking details can be added by voice, predictions can be adjusted afterward and the AI learns from corrections. It also describes dictating food weights from a kitchen scale and photographing packaged foods with its nutrition label scanner.

For nutrition data, the App Store description names the USDA database and claims 500,000 or more foods verified against USDA data. The API site describes crawling the web for nutrition values, and its documentation says the consumer app uses that same API endpoint.

The App Store nutrient list includes calories, protein, carbs, fats, fiber, sugar, sodium, cholesterol and more than 100 micronutrients, including vitamins, minerals and amino acids. SnapCalorie's download page describes calorie and macro breakdowns for every item on the plate.

Free limits and several published price lists

Up to 3 AI logs a day are free according to SnapCalorie's FAQ. Its free access page instead caps free users at one photo log a day and says they can try photo, voice, barcode and search logging. The home page calls the core features free. The two stated daily limits remain unresolved. What Premium adds is Not stated on any SnapCalorie page.

Ten purchase entries appear under SnapCalorie Premium in the App Store listing. Only Annual and Monthly include a billing period in their names.

• SnapCalorie Premium at $79.99.
• SnapCalorie Premium at $15.99.
• Annual SnapCalorie Premium at $149.00.
• Monthly SnapCalorie Premium at $19.99.
• SnapCalorie Premium at $19.99.
• SnapCalorie Premium at $49.99.
• SnapCalorie Premium at $9.99.
• SnapCalorie Premium at $99.99.
• SnapCalorie Premium at $39.99.
• SnapCalorie Premium at $59.99.

Google Play's page data gives a range of $15.99 to $149.99 per item without naming the items. SnapCalorie's checkout page offers a 7-day trial, then $149 billed yearly or $29 billed monthly. The checkout footer reads 2022. A separate older SnapCalorie page lists $199 a year and $60 a month and carries the same 2022 footer. We found no SnapCalorie Premium price on www.snapcalorie.com.

The terms say trial length is specified at account creation, then the selected monthly or yearly subscription starts automatically. They exclude refunds or credits for partial subscription periods. The subscription page directs cancellation through More, Settings, Membership Management in the app.

Is SnapCalorie accurate according to its claims and the studies?

About 15% average calorie error is the claim in SnapCalorie's FAQ. That page compares it with a stated 20% allowance for nutrition label error, about 40% error for dietitians estimating from photos and 53% for average users of other nutrition apps. These are the FAQ's comparisons. The home page claims its computer vision is twice as accurate as nutritionists.

The App Store description claims greater accuracy than other calorie trackers without giving an error figure. It points to Nutrition5k and describes testing an algorithm on 5,000 dishes with weighed ingredients. The API documentation also cites that research for a claim of twice the accuracy of professional nutritionists from an image.

Different figures appear in older SnapCalorie pages. Its older app explanation gives about 100 calories of error on a 500 calorie dish and says a single photo log will not be more accurate than weighing food, only faster. Its older accuracy page describes over 50% error from portion estimates among average users and photo logging error as low as 16.5%.

Nutrition5k used a fixed overhead depth camera

SnapCalorie's FAQ cites Nutrition5k as peer-reviewed work accepted at CVPR and says Perception Labs and Google Research co-authored it. The 2021 Nutrition5k paper lists one author at Perception Labs, with a note that his work was done while at Google.

The paper built a dataset of about 5,000 cafeteria dishes, mostly western-style food. Each item was weighed on a scale as it was added, with nutrition calculated from USDA data. On held-out test dishes, the depth-based model had 16.5% calorie error using an overhead depth camera at a fixed 35.9 cm. A model using a single color image had 26.1% calorie error.

Its human comparison used 10 simpler plates. Sixteen non-nutritionists averaged 53% error and four nutritionists averaged 41%. Nutrition5k does not name or test the SnapCalorie app.

DietAI24 compared APIs on single-food library photos

The 2025 DietAI24 paper is a peer-reviewed article in Communications Medicine. Its authors at Emory University and UC Santa Barbara built the competing DietAI24 tool. They sent 3,000 single-food photos from a US government portion image library through SnapCalorie's API. The library photos used a controlled setting and three portion sizes per food.

On those photos, mean absolute energy error per photo was 169 kcal for SnapCalorie, 168 for Foodvisor, 277 for Calorie MAMA and 47.7 for DietAI24. SnapCalorie's mean absolute errors were 8.34 grams for fat, 21 grams for carbohydrate and 4.36 grams for protein. The paper gives no food weight result for SnapCalorie.

The authors excluded SnapCalorie from their second test set, Nutrition5k, because they say it was trained on those data. They also say controlled photography may overestimate real-world performance. Their SnapCalorie test used the API and no phone depth data.

A carbohydrate abstract covered 30 prepared meals

The 2026 report in Proceedings of the Nutrition Society is a conference abstract, not a full paper. It assessed carbohydrate grams in 30 meals of known composition prepared in a lab, using LoseIT, SnapCalorie, MyFitnessPal and MacroFactor. All four apps underestimated carbohydrate on average.

The authors set a difference under 20 grams as their acceptable limit. SnapCalorie's 95% limits of agreement extended from 24.9 grams under the known amount to 3.6 grams over it, outside that limit. MacroFactor was the only app inside the limit. Separating foods on the plate made no significant difference to the error, and that held for every app.

The DietAI24 paper used no phone depth data, the carbohydrate abstract does not report whether depth or portion features were used, and Nutrition5k did not test the SnapCalorie app. None reports a test of the phone depth scan SnapCalorie describes.

Depth claims and food databases across photo apps

TrueCal's Cal AI review reports that Cal AI also describes a phone depth sensor for food volume. The Cal AI accuracy page reports no named nutrition database in the Cal AI pages it checked. The Cal AI alternatives page reports that Cal AI calls its photo numbers approximations.

The TrueCal Foodvisor review describes food names and quantity estimates as suggestions to review and reports no named nutrition source in the checked Foodvisor pages. In the DietAI24 API test, SnapCalorie and Foodvisor were 1 kcal apart in average energy error per photo. TrueCal's calorie tracker accuracy comparison also reports no named nutrition provider on the checked Cal AI and Foodvisor home pages.

The TrueCal Cronometer review reports Cronometer's claim that photo logging matches foods to its existing data sources instead of generating nutrition values. The TrueCal MacroFactor review reports that its photo AI prefers database foods and generates entries when needed or otherwise preferable. SnapCalorie's App Store description pairs its depth measurement claim with explicitly named USDA nutrition data.

TrueCal's AI calorie tracker comparison covers photo apps. Its guide to calorie trackers with sources reports that Cal AI's home page, FAQ and terms do not explain what source information a person sees for a logged food.

TrueCal's free entry and Pro logging

In TrueCal, recording calories by hand is part of the free plan. Macros and logging through a typed description, voice, photo or barcode require Pro, as described in the product reference. Barcode entries use nutrition databases, while label photos read the package.

TrueCal requires each nutrition number to come from a source or the user. Its named sources include USDA FoodData Central, FatSecret, Open Food Facts and published restaurant and brand information. Results are estimates and can be wrong. The TrueCal app comparisons include TrueCal as one option.

SnapCalorie questions on access and evidence