How ChatGPT estimates calories
It matches your description to “typical” foods, uses assumptions for missing details, then calculates an estimate. If you say “burger and fries,” it has to guess the size, cooking method, and add-ons. If you say “cheeseburger, medium fries, ate half the fries,” the estimate gets more stable.
The prompt template (copy-paste)
Estimate calories and macros (protein, carbs, fat). Meal: [what I ate] Portions: [rough amounts] Details: [brand/recipe/cooking method, sauces, oils, drinks] Output: 1) Total calories + macros 2) Assumptions you made 3) 1-2 questions that would improve accuracy
For a deeper guide (steps, prompts, accuracy, privacy), start here: ChatGPT calorie tracker.
Examples (good vs vague)
Vague
Meal: pasta
This forces assumptions: type, sauce, portion, cheese, oil.
Better
Meal: spaghetti with meat sauce Portions: ~2 cups Details: 80/20 beef, 1 tbsp olive oil, parmesan on top
Now the big calorie drivers are explicit.
Restaurant
Meal: cheeseburger + fries Portions: 1 burger, medium fries Details: no mayo, ketchup, ate about 2/3 fries
“How much I ate” is often the difference between useful and noisy.
Common mistakes
- Forgetting drinks: lattes, smoothies, alcohol.
- Missing oils and sauces: dressing, mayo, butter, cooking oil.
- Generic packaged foods: add the brand for better estimates.
- Trying to be “exact”: ask for a range when unsure.
Accuracy expectations
No calorie tracker is perfect. The goal is consistent, useful estimates. Read: accuracy: what’s realistic.
Want the low-friction version?
TrueCal is built around the conversational workflow: you describe the meal, it estimates calories and macros, and it keeps a clean log over time. Learn more on how it works or check pricing.