Which Calorie Apps ChatGPT Recommends, by Country
ChatGPT 
Claude 
AI Mode 
Perplexity We asked ChatGPT the same 50 calorie-app and buying questions in English, German, French and Spanish, with its web search set to the United States, Germany, France and Spain. This page lists the apps it recommended in each country and the pages it cited to back them up. NutriScan built the question panel to track its own visibility, and NutriScan appears in the answers, so the disclosure section gives its exact counts.
Key findings
All figures below come from the free-tier model (gpt-6-luna), 50 answers per country.
| # | Finding |
|---|---|
| 1 | In the United States, ChatGPT recommended Cronometer in 19 of 50 answers (38%) and called it the top pick in 11 (run of 2026-10-31). |
| 2 | In Germany, ChatGPT recommended Yazio in 25 of 50 answers (50%) and FDDB in 18 (36%). FDDB did not appear in a single US or French answer (run of 2026-10-31). |
| 3 | In France, Yazio led with 18 of 50 answers, followed by FatSecret with 11 and Foodvisor with 10 (run of 2026-10-21). In Spain, FatSecret led with 14, Yazio 13 and Cronometer 12 (run of 2026-10-29). |
| 4 | MyFitnessPal was recommended in 7 of 50 US answers, against 19 for Cronometer and 9 for Lose It! (run of 2026-10-31). |
| 5 | Help and support pages made up 40% of the links ChatGPT cited in US answers (46 of 114) and 35% in German answers (42 of 119), against 16% in French answers (19 of 120) and 21% in Spanish answers (24 of 114). |
Top 10 apps, by country
"Recommended" counts the answers whose list of suggested apps includes the app. "Top pick" counts the answers that name it as the first choice. Free is gpt-6-luna, which ChatGPT serves on its Free and Go plans. Paid is gpt-6-sol, which it serves on Plus and above. Both columns are out of 50 answers.
United States (2026-10-31)
| Rank | App | Recommended, free | Top pick, free | Recommended, paid |
|---|---|---|---|---|
| 1 | Cronometer | 19 (38%) | 11 | 20 |
| 2 | Lose It! | 9 (18%) | 2 | 4 |
| 3 | MyNetDiary | 8 (16%) | 6 | 10 |
| 4 | MyFitnessPal | 7 (14%) | 1 | 9 |
| 5 | MacroFactor | 7 (14%) | 5 | 5 |
| 6 | SnapCalorie | 4 (8%) | 4 | 7 |
| 7 | Caltrac | 2 (4%) | 1 | 0 |
| 8 | Healthify | 2 (4%) | 1 | 2 |
| 9 | Eat This Much | 2 (4%) | 2 | 2 |
| 10 | PCOS Polly | 2 (4%) | 1 | 0 |
Germany (2026-10-31)
| Rank | App | Recommended, free | Top pick, free | Recommended, paid |
|---|---|---|---|---|
| 1 | Yazio | 25 (50%) | 10 | 25 |
| 2 | FDDB | 18 (36%) | 14 | 16 |
| 3 | Lifesum | 5 (10%) | 0 | 6 |
| 4 | Cronometer | 5 (10%) | 0 | 9 |
| 5 | Linup | 4 (8%) | 4 | 4 |
| 6 | Foodvisor | 4 (8%) | 0 | 1 |
| 7 | FatSecret | 4 (8%) | 2 | 5 |
| 8 | Peach | 2 (4%) | 1 | 0 |
| 9 | MIMBA | 2 (4%) | 1 | 1 |
| 10 | EAT SMARTER | 2 (4%) | 0 | 0 |
France (2026-10-21)
| Rank | App | Recommended, free | Top pick, free | Recommended, paid |
|---|---|---|---|---|
| 1 | Yazio | 18 (36%) | 9 | 19 |
| 2 | FatSecret | 11 (22%) | 9 | 14 |
| 3 | Foodvisor | 10 (20%) | 7 | 5 |
| 4 | Cronometer | 7 (14%) | 1 | 6 |
| 5 | MyFitnessPal | 3 (6%) | 0 | 4 |
| 6 | Open Food Facts | 3 (6%) | 0 | 3 |
| 7 | SnapCalorie | 3 (6%) | 2 | 1 |
| 8 | MacroFactor | 2 (4%) | 0 | 1 |
| 9 | mySugr | 2 (4%) | 1 | 2 |
| 10 | BeMum | 2 (4%) | 0 | 0 |
Spain (2026-10-29)
| Rank | App | Recommended, free | Top pick, free | Recommended, paid |
|---|---|---|---|---|
| 1 | FatSecret | 14 (28%) | 11 | 13 |
| 2 | Yazio | 13 (26%) | 8 | 11 |
| 3 | Cronometer | 12 (24%) | 3 | 16 |
| 4 | MyFitnessPal | 4 (8%) | 1 | 6 |
| 5 | Fotocal | 3 (6%) | 1 | 0 |
| 6 | MisCalorías | 3 (6%) | 1 | 1 |
| 7 | Foodvisor | 3 (6%) | 0 | 2 |
| 8 | mySugr | 2 (4%) | 0 | 1 |
| 9 | Diabetes a la Carta | 2 (4%) | 2 | 2 |
| 10 | vitaloop | 2 (4%) | 0 | 1 |
In every country, 49 of 50 answers recommended at least one app. The one answer without a list in each country replied to the question on how accurate photo calorie counters are.
What ChatGPT cites
Each answer comes with the links ChatGPT used. We sorted every cited link into four types.
| Country | Links cited | Help and support pages | App makers' own pages | App Store and Google Play | Publishers, health bodies, research |
|---|---|---|---|---|---|
| United States | 114 | 46 (40%) | 29 (25%) | 12 (11%) | 27 (24%) |
| Germany | 119 | 42 (35%) | 28 (24%) | 13 (11%) | 36 (30%) |
| France | 120 | 19 (16%) | 50 (42%) | 25 (21%) | 26 (22%) |
| Spain | 114 | 24 (21%) | 43 (38%) | 17 (15%) | 30 (26%) |
Most help pages belong to the apps themselves. In the US run, the most cited were Lose It!'s Zendesk help centre (9 links), Cronometer support (9), MyNetDiary help (7), MacroFactor help (5) and the MyFitnessPal help centre (5). In the German run, FDDB's help centre alone supplied 15 links and Yazio's 14.
How stable the picks are
The panel runs the same questions every few days. Below is every full run with the location set (free model, 50 answers each). The US, German and French leaders held in every run. In Spain, Yazio led the first run and FatSecret the next two.
| Country | Run | Top 3 recommended | Help pages share |
|---|---|---|---|
| United States | 2026-10-14 | Cronometer 20, Lose It! 11, MacroFactor 9 | 27% |
| United States | 2026-10-19 | Cronometer 20, MacroFactor 8, MyFitnessPal 8 | 26% |
| United States | 2026-10-31 | Cronometer 19, Lose It! 9, MyNetDiary 8 | 40% |
| Germany | 2026-10-14 | Yazio 23, FDDB 19, Lifesum 7 | 40% |
| Germany | 2026-10-20 | Yazio 25, FDDB 18, Cronometer 8 | 37% |
| Germany | 2026-10-31 | Yazio 25, FDDB 18, Lifesum 5 | 35% |
| France | 2026-10-14 | Yazio 18, Foodvisor 13, FatSecret 11 | 17% |
| France | 2026-10-21 | Yazio 18, FatSecret 11, Foodvisor 10 | 16% |
| Spain | 2026-10-15 | Yazio 18, FatSecret 13, Cronometer 11 | 22% |
| Spain | 2026-10-22 | FatSecret 13, Cronometer 12, Yazio 12 | 12% |
| Spain | 2026-10-29 | FatSecret 14, Yazio 13, Cronometer 12 | 21% |
The app rankings moved by a few answers between runs. The share of help pages moved more: the US share went from 26% to 40% between October 19 and October 31, so treat that figure as one snapshot.
What changes between settings
| Setting | What we ran | What changed |
|---|---|---|
| Paid model | gpt-6-sol on the same questions, same day | The leader stayed the same in the US (Cronometer 20), Germany (Yazio 25) and France (Yazio 19). In Spain, Cronometer moved first with 16, ahead of FatSecret with 13. |
| No location | gpt-6-luna and gpt-6-sol on 2026-10-14 with no country set, questions still in the local language | German and French answers moved toward US apps. In German, Cronometer rose from 5 to 14 recommendations against the located run of the same day, and FDDB fell from 19 to 11. In French, Cronometer led with 16 and Yazio fell from 18 to 13. The US list barely changed. |
| Smaller model | gpt-5.4-mini on 2026-10-14, the 20 app questions only, no location | MyFitnessPal appeared in 11 of 20 US answers, 10 of 20 German answers and 11 of 20 French answers. Yazio appeared in 3 German answers and FDDB in 1. This run has no extracted pick lists, so the counts come from app names found in the answer text. |
The dataset covers the four headline runs. The other runs are summarized in the two tables above.
Methodology
| Item | Detail |
|---|---|
| Question set | 50 questions per language: 20 direct app questions ("What is the best AI calorie tracker app?") and 30 buying questions written the way people ask them ("suggest me a good calorie tracker app where i can just take a photo of my food, i hate typing everything in. free or cheap"). The German, French and Spanish sets are adapted to the market rather than translated word for word: the German wedding question names Yazio as the app the person gave up on, where the English one names MyFitnessPal. The panel also holds need, fact and hub questions, which are left out here. |
| Models | gpt-6-luna (served on ChatGPT Free and Go) and gpt-6-sol (served on Plus and above), called through the OpenAI Responses API with the web search tool. Smaller-model check: gpt-5.4-mini-2026-03-17. |
| Location | Web search user_location set to the country: US, DE, FR or ES. The no-location runs leave it unset. |
| Headline runs | United States 2026-10-31, Germany 2026-10-31, France 2026-10-21, Spain 2026-10-29. Earlier runs from 2026-10-14 to 2026-10-22 are used for the stability table. 17 run files in total. |
| Pick extraction | A second gpt-6-luna call reads each answer and returns the ordered list of apps it recommends and its top pick (the one the answer calls its first choice, else the first listed). App names are merged case-insensitively, with plan names dropped (Yazio Pro counts as Yazio, MyFitnessPal Premium as MyFitnessPal) and spellings merged (Lose It and Lose It!, Fddb and FDDB, HealthifyMe and Healthify). Each app counts once per answer. |
| Help page | A cited link whose host or path contains "help", "support" or "faq", or whose path contains "hc/" (Zendesk help centres). App store links are apps.apple.com and play.google.com. An app maker's own page is a link on a domain that carries the name of an app recommended in the panel. Everything else counts as publishers, health bodies and research. |
| Limits | These are API answers from one vendor's models, not screenshots of the ChatGPT app, which can add memory, personalization and a different search setup. Each run is a snapshot of one day, and the runs are a few days apart, so they are not a time series. Answers vary between calls, and each question was asked once per run and setting. 50 answers per country is a small sample: one answer moves a share by 2 points. The question set leans toward questions NutriScan wants to rank for, such as AI photo logging, voice coaching and health conditions. 8 or 9 of the 50 questions per language name an app, mostly as one the person wants to replace. None of those answers recommended the app the question named, so Yazio's 25 German recommendations all come from the 47 questions that do not mention it. |
| Conflict of interest | NutriScan makes a calorie tracking app, built this panel to measure its own visibility and appears in the answers. Its counts are below. |
Disclosure: NutriScan in these answers
NutriScan is one of the apps the panel tracks. In the headline runs it was recommended in these answers, out of 50 per country:
| Country | Run | Recommended, free | Top pick, free | Recommended, paid | Top pick, paid |
|---|---|---|---|---|---|
| United States | 2026-10-31 | 1 | 1 | 1 | 1 |
| Germany | 2026-10-31 | 1 | 0 | 1 | 1 |
| France | 2026-10-21 | 1 | 1 | 3 | 2 |
| Spain | 2026-10-29 | 1 | 1 | 1 | 1 |
In every country the free-model mention came from the same question: whether an app offers an AI nutritionist you can talk to by voice. The two extra French mentions on the paid model came from a PCOS question and a question about a Yuka-style photo check for whole meals. Across all earlier runs NutriScan was named in 0 to 3 answers per country and setting, and in none of the gpt-5.4-mini answers. NutriScan is left out of the charts' top six because it never ranked that high.
Dataset
Download the per-question picks (CSV). It has one row per answer in the headline runs (400 rows): run date, country, location setting, model, question ID and text, the apps recommended in order, the top pick, the number of links cited and the number of help pages among them. NutriScan releases the table under Creative Commons Attribution 4.0 (CC BY 4.0); reuse it with credit and a link to this page.
To cite this page: NutriScan, "Which Calorie Apps ChatGPT Recommends, by Country", nutriscan.app/research/ai-assistant-app-recommendations, published 2026-10-31. The charts carry a credit line and can be reused with a link back.
Related research: Fast-food calories by country.
Where NutriScan fits
NutriScan works from a photo: you photograph the meal, check the items and portions it finds, and get calories, protein, carbs and fat. Each meal also gets an A to E meal grade, so you can compare two meals at a glance. You can also log a meal by voice, and on plans that include Monika, the in-app AI nutrition coach, you can ask her questions in a voice call. The free plan includes 5 AI scans a week. Premium costs about $59 a year and removes the weekly limit. NutriScan has 150k+ downloads and a 4.6-star rating.