AI Calorie
Photograph a meal or scan a barcode to log calories and macros locally.
AI: publik API or your key
not yet reviewed by publik. It is live now; the founder has not reviewed this version yet.
by 
A
AI Calorie
open source · on publik
Where AI Calorie’s AI runs, and what it costs
Prices are for one typical use: one request of about 1,500 words sent and 375 words back. A higher quality score is better.
On your computer
Not checked yet: publik has not read whether AI Calorie can run its AI on your computer.
publik API
- Price
- $0.0010 per use, $1.00 per 1,000 usespublik-fast, the level AI Calorie uses most
- Quality
- GLM-5.3 Flash: 42
- Setup
- Built in. No key to paste; linking your publik account gives $0.05 of free use, once.
Your own key
- Price
- $0.0006 per use, $0.55 per 1,000 usesGLM-5.3 Flash at OpenRouter’s list price, before its fee for buying usage
- Quality
- GLM-5.3 Flash: 42
- Setup
- Open a provider account, add a card, paste the key into AI Calorie.
Quality and price, side by side
| Model | Quality | Per 1,000 uses |
|---|---|---|
| On your computer (Ollama, 4-bit download size) | ||
| Granite 4.2 3B2.2 GB | 9 | $0 |
| Phi-4 Mini2.5 GB | 6 | $0 |
| Llama 3.1 8B4.9 GB | 7 | $0 |
| gpt-oss 20B14 GB | 9 | $0 |
| Gemma 4 31B20 GB | 19* | $0 |
| Qwen3.5 35B-A3B24 GB | 19* | $0 |
| publik API | ||
| publik-fastGLM-5.3 Flash · AI Calorie | 42 | $1.00 |
| publik-balancedMiMo-V2.6-Pro | 46 | $10.00 |
| publik-smartGPT-6 Sol | 48 | $18.00 |
Quality: Artificial Analysis Intelligence Index v4.3.2, read 2026-09-22 (publik API models 2026-09-25); * = estimated by Artificial Analysis. Sizes: the Ollama library, read 2026-09-22.
What you pay for
- On your computer: nothing per use. You pay in disk space, memory and electricity, at lower quality.
- publik API: publik’s published price for each use, in dollars, from your publik balance. It is above the model’s cost; the difference runs publik and pays the app’s builder.
- Your own key: the provider’s price, billed to an account you keep with the provider.
How to install AI Calorie
Every step written out. No terminal experience needed. Pick your setup.
README
Open in GitHubAI Calorie
Photograph a meal, get its calories and macros.
AI Calorie is a local web app. It runs a small Node server on your machine, serves
the built front end from it, and keeps your log in a JSON file under
~/.ai-calorie/. Nothing about your eating leaves the machine except the meal
photo you choose to have analysed.
What it does
- Scan a meal — take a photo (or pick one, or just describe it in words) and get calories, protein, carbs, fat, fibre, sugar, a 0–100 health score and an ingredient-by-ingredient breakdown.
- Scan a package — point the camera at a barcode and the product is looked up in Open Food Facts, so the figures are the label's own rather than a guess. For anything not in the database, photograph the nutrition panel: it is read on your own machine, so it costs nothing and the photo is never uploaded.
- Correct it in words — "it was a double portion", "no dressing" — and the estimate is recomputed.
- Daily targets — built from your height, weight, age, activity and goal with Mifflin–St Jeor plus an activity factor; editable afterwards.
- Today — a calories-remaining ring, macro bars, a logging streak, and your meals grouped by breakfast / lunch / dinner / snacks.
- History — every meal you have logged, searchable, grouped by day.
- Progress — average calories and protein, days logged, a daily calorie chart against your target, and a weight trend.
Running it
npm install
npm start
Then open http://localhost:4273. The first npm start builds the interface,
which takes a second; later starts skip it. Node 20 or newer.
For development with hot reload:
npm run dev
Where the numbers come from
- Barcodes — Open Food Facts, open data, no key and no account. Figures are the manufacturer's, scaled to one serving. Free.
- Nutrition panels — read on your machine with a WebAssembly build of Tesseract, then cross-checked (sugars cannot exceed carbohydrate, the macros must account for the stated energy). Free, offline after first use, and the photo never leaves the machine. If too little comes out legibly, the app offers the AI read below instead.
- Meal photos and written descriptions — a vision model. This is the only path that costs anything.
Meal photos resolve a provider in this order:
- Your own key —
ANTHROPIC_API_KEY, if you set one. Get it from https://console.anthropic.com/settings/keys. - publik API — a released build carries
publik-app-token.txtand provisions a per-install key on first launch, after showing you what that means and asking. No key to paste. See https://publikhq.com/developers.
Copy .env.example to .env only if you want to override that. With no provider
set up, barcode and panel scanning still work — they never needed one — and meal
photos ask you to choose a provider first.
Self-test
A release build answers publik's startup self-test:
PUBLIK_SELF_TEST=1 npm start
It makes one real call through publik API and prints a single line
(publik self-test: ok <slug> <version>), then exits.
Accuracy
These are estimates from a photo, not measurements. Portion size is the largest source of error — check it on the result screen and correct it when it looks off. AI Calorie is not medical advice.
Licence
MIT — see LICENSE. The numbers this app produces are estimates, and the licence's "as is, without warranty" terms apply to them too.