Hero Ambient Glow Background
The API for AI Calorie & Meal-Tracking Apps

Clinical-grade nutrition intelligence. Verified down to the calorie.

Deconstruct meal photos, barcode scans, and natural language into structured USDA nutrition JSON in <18ms with a developer-friendly REST API.

200 free requests / month • Full multimodal access • No credit card required

Live Multi-Modal Meal Decomposition

Click a plate photo to inspect real ingredient segmentation and verified macros

Grilled Salmon Quinoa Bowl
Vision Scan
14ms Edge KV
Grilled Salmon Quinoa Bowl
High Protein • Pescatarian
Calories
520
Protein
42g
Carbs
38g
Fat
22g
Decomposed Ingredients (4)grounded_via_usda
1
Wild Alaskan Salmon Fillet
180g grilled
260 kcal
2
Cooked White Quinoa
120g
140 kcal
3
Hass Avocado Slices
50g (1/3 medium)
80 kcal
4
Steamed Edamame & Broccoli
80g
40 kcal
4(42) + 4(38) + 9(22) = 518 kcal (0.4% variance)
API Reference
4M+ Foods
Grounded in USDA & Barcode DBs
< 18ms
Anycast Edge KV Latency
99.4% Pass
USDA Ground Truth Match
0.4% Var
4-4-9 Atwater Math Consistency
Developer Experience

Stop Stitching 5 Brittle APIs Together

Building an AI calorie or fitness app shouldn't require balancing raw vision models, ungrounded scrapers, and hallucinated macro math.

The DIY StackHigh Lag & Hallucinations
  • ✕Generic Vision LLM: $20–$35/1k calls with ~2,500ms latency
  • ✕Separate Barcode API: Inconsistent GTINs and extra SaaS fees
  • ✕USDA DB Scrapers: Fragile 100g conversion & uncalibrated portions
  • ✕Impossible Calorie Math: 12–15% thermodynamic violations
Result: Weeks of boilerplate & user complaints over wrong macros
The FoodVivo API1 Unified REST Endpoint
  • ✓Multi-Modal Vision: Plate deconstruction & label OCR in <18ms
  • ✓Integrated 4M+ Barcodes: Dual-source GTIN + branded food resolution
  • ✓4-4-9 Mathematical Proof: 100% physically consistent macros
  • ✓Predictable Unit Economics: From $1.31/1k calls with 200 free req/mo
Simple HTTP integration • Ship in 1 afternoonTest live

Accuracy is the only moat in food intelligence.

When users log meals, impossible calorie math and hallucinated portions destroy trust. FoodVivo enforces the mathematical and biochemical guarantees that raw LLMs lack.

4-4-9 Thermodynamic Verification

Every prediction is constrained by physical chemistry: Calories = (4×Protein) + (4×Carbs) + (9×Fat). Any prediction violating caloric equilibrium is mathematically recalibrated.

0.4% Max Mathematical Variance

USDA Ground-Truth Anchor

Predictions are mapped directly to USDA FoodData Central (Foundation Foods & SR Legacy) biochemical assays rather than unverified web approximations.

99.4% Biochemical Alignment

Sub-18ms Anycast Edge Speed

Common foods, recipes, and 4M+ barcodes resolve at the edge across 300+ global data centers with sub-18ms latency — eliminating 3-second camera lag.

Sub-18ms Tier-0 Response
Architecture

How the pipeline works

A 4-stage pipeline that handles multimodal intake, grounds ingredients against USDA databases, and verifies caloric math.

STAGE 01

Multimodal Ingestion

Accepts text descriptions, plate photos, UPC/EAN barcodes, or packaging labels through a single REST API endpoint.

Text • Photo • Barcode • OCR
STAGE 02

Neural Grounding

Cross-references USDA FoodData Central and verified brand databases to eliminate hallucinations on regional and restaurant dishes.

USDA Ground Truth
STAGE 03

Serving-Size Arbiter

Normalizes raw 100g database rows into real-world packaging units (e.g. '2 scoops (35g)' or '1 can (330ml)').

Packaging Units
STAGE 04

Mathematical Validation

Verifies caloric consistency (4P + 4C + 9F ≈ Calories) and caches verified results at the edge for sub-18ms responses.

< 18ms Edge Cache
Developer Integration

Drop-in integration

Start logging meals in minutes with clean REST API examples for any language.

POST/v1/food/analyze
const response = await fetch("https://foodvivo-api.flappybird-assets.workers.dev/v1/food/analyze", {
  method: "POST",
  headers: {
    "Authorization": "Bearer sk_live_your_api_key",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: "2 scrambled eggs with 1 slice sourdough toast and half avocado",
    context: "Breakfast",
  }),
});

const { data, meta } = await response.json();
console.log("Meal Title:", data.meal_title);
console.log("Totals:", data.totals);
console.log("Macro Split:", data.macro_split);
Continuous Ground Truth Evals

99.4% Accuracy Across 25+ Golden Culinary Scenarios

Evaluated continuously against USDA FoodData Central ground truth across single ingredients, complex restaurant dishes, global cuisines, and packaged barcodes.

< 0.8% Calorie Deviation•100% 4-4-9 Math Consistency