What is the best free calorie counter app?
3 of 4 assistants converge on Cronometer. Most frequent first pick: Cronometer (8 of 12).
November 2026 round · asked November 1, 2026
Each month we ask ChatGPT, Claude, Gemini and Perplexity the questions people really type, three times each, and publish every answer as it came back, plus the data. This round: 14 questions, 168 answers.
At least three of the four assistants named the same product first.
3 of 4 assistants converge on Cronometer. Most frequent first pick: Cronometer (8 of 12).
4 of 4 assistants converge on Zero. Most frequent first pick: Zero (11 of 12).
3 of 4 assistants converge on Strava. Most frequent first pick: Strava (9 of 12).
4 of 4 assistants converge on Sleep Cycle. Most frequent first pick: Sleep Cycle (12 of 12).
3 of 4 assistants converge on 1Password. Most frequent first pick: 1Password (8 of 12).
3 of 4 assistants converge on Babbel. Most frequent first pick: Babbel (8 of 12).
4 of 4 assistants converge on Todoist. Most frequent first pick: Todoist (12 of 12).
No product was the first choice of three or more assistants. These are the questions where asking a second AI is most likely to change your answer.
No convergence: the assistants each pick differently. Most frequent first pick: Cronometer (5 of 12).
No convergence: the assistants each pick differently. Most frequent first pick: Cal AI (3 of 12).
Split: two assistants agree, two go their own way. Most frequent first pick: Nike Training Club (8 of 12).
Split: two assistants agree, two go their own way. Most frequent first pick: Insight Timer (7 of 12).
Split: two assistants agree, two go their own way. Most frequent first pick: YNAB (8 of 12).
Split: two assistants agree, two go their own way. Most frequent first pick: OneNote (7 of 12).
Split: two assistants agree, two go their own way. Most frequent first pick: Streaks (7 of 12).
More and more people ask an assistant instead of a search engine when they want a recommendation. The answer they get depends on which assistant they ask, and often on when they ask it. We put the same plain question to all four, with web search on, and show where the answers meet and where they don't.
We don't add our own pick, and we don't edit the answers. If an assistant says something wrong, that stays in the record. Read how we ask, or browse every question.