Can ChatGPT track calories?

ChatGPT can help draft a calorie and macro estimate when you give it enough meal detail, but it should not be treated as a measured food log. Use it as a first pass. Ask it to list assumptions, show a range when portions are uncertain, and name the details that could change the estimate.

The practical goal is not to make ChatGPT sound confident. The goal is to make the answer easier to review before you save it in a food log or rewrite it in Mori.

Why a better prompt helps

A vague prompt such as "how many calories are in this meal?" asks the model to fill in everything you left out: portion, cooking fat, sauce, side dishes, drink, brand, restaurant recipe, and whether the plate was finished. Those guesses can create a clean-looking answer that hides the messy parts.

A better prompt separates known details from unknown ones. It tells the assistant to use ranges for uncertain portions, avoid medical advice, and explain which assumption matters most. That makes the answer slower than a magic number, but far more useful.

This is also where Mori fits. Mori is built around editable estimates from plain-language meal descriptions. You can use the prompt builder to sharpen the meal details, then log the final description in Mori and correct the calorie, protein, carb, and fat fields as better information appears.

What current evidence says about AI calorie estimates

The fresh Last30Days run found recent YouTube examples where creators are building or replacing calorie trackers with ChatGPT or Claude. That shows real user interest in chat-based food logging, but the evidence was thin: Reddit results were noisy, YouTube transcript capture was degraded, and X was unavailable in the local setup.

Current research supports a cautious workflow. A 2025 study in Nutrients found that ChatGPT-4 identified foods in meal photographs well, but performed poorly for medium and large portion weights and accurate nutrient estimates. That pattern matches the everyday problem: the model may name the food correctly while still missing how much food is there.

Research presented at NUTRITION 2026, summarized by the American Society for Nutrition, tested four photo-based apps against meals prepared in a controlled metabolic kitchen. The apps underestimated calories and fat on average, and the report notes that the findings were conference results rather than a fully peer-reviewed journal article. Treat that as a warning signal, not a final universal number for every app or meal.

How to use the prompt builder

Fill in the meal, source, portion, main ingredients, cooking fat, extras, and any known label or menu data. Copy the prompt into ChatGPT, Claude, Gemini, or another assistant. If the answer gives one exact number for a vague meal, ask for the assumption list and a range before saving it.

When the meal is packaged, the label should win over the chatbot. When the meal is from a chain restaurant, the restaurant nutrition page should win when it matches your order. When the meal is homemade and the recipe matters, measured ingredients are stronger than a photo or a memory.

If the prompt returns calories and macros, run the result through the AI calorie estimate checker. It can compare the calorie number with macro-derived calories and remind you to check portions, oil, sauces, toppings, drinks, labels, and source quality.

How this becomes a Mori entry

Use the prompt output to write a clean meal description. A good entry might read: "Restaurant grilled chicken rice bowl, about 2 cups, chicken breast, white rice, cucumber, unknown oil, 2 tablespoons yogurt sauce, estimate range because restaurant recipe is unknown."

Then log it in Mori as an editable estimate. If the answer from ChatGPT includes assumptions that seem wrong, fix the description before accepting the numbers. If the meal has a label or restaurant listing, add that source detail. If you only know calories, log calories first and leave uncertain macros editable rather than inventing them.

For simpler input, use the AI meal description builder. For choosing whether text, voice, photo, barcode, or database search fits you better, use the calorie tracker app fit checker or the AI food tracker app for iPhone guide.

Sources used

This page uses current research and recent community evidence as guardrails. The prompt itself does not rely on one social post. It reflects the repeated pattern that calorie estimates are more reviewable when portions, cooking fat, sauces, and source data are explicit.

Limits and safety notes

This tool is educational. It does not provide medical advice, diagnose a condition, prescribe calories, or decide whether a meal is appropriate for you. A chatbot answer is not a substitute for a clinician, registered dietitian, official label, restaurant nutrition page, or measured recipe when those sources matter.

If you have diabetes, kidney disease, pregnancy-related guidance, a history of eating disorders, food allergies, celiac disease, or another clinical concern, follow your care team's instructions. If calorie tracking creates fear, guilt, binge and restriction cycles, compensatory exercise, or distress, pause the tracking workflow and seek qualified support.