The most accurate photo scanner
Users log a meal in one snap. The API returns a meal name, itemized ingredients, portions, and full macros, editable with AI follow-ups.
POST /v1/scanmulti-itemeditable
B2B APIs & white-label
Lifestyle intelligence, delivered as an API
Plug-and-play endpoints for the world's most accurate food photo scanner, a 54M verified-food database, and glucose predictions with or without a sensor. Integrate in hours, or ship the whole experience under your own brand.
Example request
POST /v1/scan
Authorization: Bearer <api_key>
{ "image": "<base64>" }
200 OK
{
"meal": "Chicken burrito bowl",
"items": 6,
"calories": 712,
"glucose_peak_mgdl": 148,
"swaps": 3
}Verified foods54M+
Prediction horizon33 hours
Time to integrateHours, not quarters
PlatformsiOS, Android, web
01Capabilities
Six endpoints, one metabolic picture
Take the whole suite or a single call. Every response is structured, versioned, and documented.
Voice and text quick logging
A plain sentence becomes a structured meal. Describe it out loud or type it, and the parser resolves foods, quantities, and preparation.
POST /v1/parsenatural language
54M verified foods, never crowd-sourced
Grocery items, branded packages, local and chain restaurant menus, and recipes, with barcode lookup across the full catalog.
GET /v1/foodsbarcoderestaurants
Glucose predictions with or without a CGM
Prototypical curves by demographics, condition, and activity for the 128M people who will never wear a sensor, or personalized curves when CGM data exists.
POST /v1/glucoseCGM-033h horizon
Personalized food swaps
Return healthier alternatives for any logged food, plus ingredient-level changes ranked by nutritional and glycemic impact.
GET /v1/swapsfood scores
CGM-enabled fasting and sleep insights
Use glucose as a proxy for eating events to derive eating windows, dinner-to-sleep gaps, and logging reminders without user friction.
GET /v1/insightseating window
02Benchmarks
State of the art on food photos
1,000 real-world food images with human-reviewed annotations, diverse cuisines, realistic lighting and camera angles. Every model scored on the same set.
RankModelOverall
1january-ai-vision86.24
2gpt-4o74.11
3gpt-4o-mini66.44
4gemini-2.5-pro60.67
5gemini-2.5-flash60.65
86%
Scan accuracy
Across ingredient recognition and macronutrient estimation, more than 12 accuracy points ahead of the next best app.
+12.1
Points over GPT-4o
Overall composite score of 86.24 versus 74.11 for GPT-4o on the same 1,000 images.
1,000
Benchmark images
Real meals from the January app, each with human-validated meal names, ingredients, and macros.
Lowest
Macro wMAPE
Weighted mean absolute percentage error on calories, carbs, protein, and fat, where lower is better.
Methodology
- Ground truth from reputable public datasets plus human-reviewed January app annotations.
- Each photo run twice per app, best of two kept, processing time recorded, failures marked when both attempts fail.
- Ingredient accuracy is the share of correctly identified ingredients against ground truth, reported as precision, recall, and F1.
- Macro accuracy averages the Mean Absolute Percentage Error of each macro against ground truth values.
- Overall Score is a 0 to 100 composite of meal name similarity, ingredient F1, macro wMAPE, and response time.
Open and reproducible
The paper, the code, and the dataset are public, so you can rerun every number yourself before you sign anything.
03White-label
Ship the whole app under your brand
Every feature in the January app, restyled as yours: your logo, your palette, your onboarding, your store listing. We keep the models current; you keep the relationship with your users.
Your brand, end to end
Themeable UI, copy, and assets across iOS, Android, and web.
Design support included
A Figma content package plus custom design help from our team.
A dedicated Slack channel
Direct line to the engineers who maintain the models.
Compliance-ready
HIPAA-aligned data handling, audit trails, and regional hosting options.
04Integration
Live in three steps
01
Request a key
Sign up in the docs and get sandbox credentials the same day.
02
Call the endpoints
REST and JSON, with client SDKs for iOS, Android, and TypeScript.
03
Go to production
Move to production keys, then scale with usage-based limits and support.
05Pricing
Three tiers
Plans start at $1,499 / month
Tier
Core
The world's best food search and logging.
Photo scanning
Barcode search
Voice & text search
54M food database
Tier
Intelligence
Add glucose predictions and nutritional intelligence.
Everything in Core
CGM-0 glucose predictions
Personalized food swaps
Food scores
Tier
Precision
Add CGM-enabled AI insights.
Everything in Intelligence
Precision glucose predictions
Fasting & sleep insights
CGM-enabled logging reminders
All plans include a dedicated Slack channel, a Figma content package, and custom design support.
Get started
Request an API key and start testing today.
Or see the platform behind the endpoints on the January AI Platform page.