ALURE
Accuracy benchmark

How accurate is Alure?

We ran Alure's food scan and MyFitnessPal over the same 52 foods, then scored both against independent USDA and official references. Neither tool is treated as the correct answer. Every number is below.

Last updated June 11, 2026

How to read this

This benchmark is automated and reproducible. Alure's numbers come straight from the app's real food-scan pipeline. MyFitnessPal's come from searching each food in its web app and taking the top result, the same thing a real user sees.

Both tools are scored by the same formula against the same independent references, and the scoring never knows which tool it is grading. We held Alure to exactly the same test as MyFitnessPal.

ALURE
84/100
42 of 52 passed
MYFITNESSPAL
66/100
23 of 52 passed
Test Calories Score / 100
Ref Alure MFP Alure MFP
Simple Single-ingredient foods at a common portion, with exact USDA reference values. 87 67
1 medium banana 105 105 105 79 77
2 large eggs 143 143 140 93 77
1 cup cooked white rice 205 234 200 75 80
1 medium avocado 240 240 100 100 26
6 oz grilled chicken breast 280 281 200 97 61
1 cup whole milk 149 149 150 100 100
1 medium baked sweet potato 103 130 26 57 28
1 cup plain Greek yogurt 179 130 70 67 40
1 tablespoon olive oil 119 123 119 98 99
100g raw salmon fillet 208 208 182 100 84
Portions Foods given with an explicit weight or measure. Tests accuracy when the serving size is stated, not guessed. 94 63
200g chicken breast, grilled 330 330 474 100 39
50g rolled oats, dry 190 190 188 92 89
150g cooked brown rice 166 165 263 94 46
30g almonds 174 175 160 97 93
1/2 cup peanut butter 753 717 200 94 33
85g dark chocolate 70% 510 493 168 89 32
120g cooked pasta 188 190 157 97 85
45g whey protein powder 173 180 180 89 88
Homemade Multi-ingredient home-cooked meals. References are USDA composites for a typical preparation; wider tolerance reflects natural variation. 82 65
grilled cheese sandwich 400 331 200 79 63
scrambled eggs with toast and bacon 400 364 336 78 66
spaghetti with meat sauce 440 553 329 75 78
chicken stir fry with rice 485 509 277 87 23
turkey and cheese sandwich 360 402 360 77 83
bowl of oatmeal with banana and honey 335 335 320 91 53
caesar salad with grilled chicken 450 468 180 76 43
peanut butter and jelly sandwich 400 377 373 85 94
tuna salad sandwich 385 420 217 85 55
pancakes with butter and maple syrup 530 496 560 88 96
Restaurant Chain-restaurant items with officially published nutrition data. 86 92
McDonald's Big Mac 590 590 550 100 95
Chick-fil-A original chicken sandwich 440 460 440 96 99
Chipotle chicken burrito bowl 655 630 510 94 71
Starbucks grande caramel frappuccino 380 370 410 90 96
Subway 6 inch turkey breast sub 280 470 370 42 72
Taco Bell crunchy taco 170 170 170 89 98
Wendy's Dave's Single 590 570 590 96 99
Panda Express orange chicken entree 490 420 490 87 98
In-N-Out Double-Double 610 670 670 90 89
large pepperoni pizza slice from Domino's 300 380 300 74 100
International Non-US dishes commonly ordered out. References are composites of typical restaurant preparations; the widest tolerance reflects regional variation. 67 56
chicken tikka masala with naan bread 750 580 530 77 73
beef pho 450 459 638 75 50
tteokbokki 330 685 420 6 72
mapo tofu with rice 560 484 645 85 64
jollof rice with chicken 550 432 180 63 29
chicken shawarma plate with rice 700 597 353 84 55
pork adobo with rice 520 584 110 81 25
khao soi 600 834 666 63 83
Typos Inputs deliberately misspelled to test how each tool copes with real-world typing. References are for the correctly-spelled food. 86 39
chiken parmasan 570 548 60 75 10
mackdonal chiken nugits 10 piece 440 420 270 96 62
starbux vanila latte grande 250 290 70 83 29
avacado tost with egg 350 362 258 92 55
2 slise peporoni piza 596 540 988 89 31
greak yogert with hunny 200 194 100 81 48

Each score blends calories (40%), protein (25%), carbs (20%) and fat (15%) against the reference. Chip colour is a guide: green is a close read, amber is in the ballpark, rose is well off.

Methodology

Reference sources

  • USDA FoodData Central — the authoritative nutrient database, scaled to each stated portion, for single foods and precise weights.
  • Official restaurant nutrition — published figures from each chain for their menu items.
  • Documented composites — USDA composites of a typical preparation for home-cooked and international dishes, which carry a wider tolerance because the same name spans a real range of recipes.

How each tool was run

  • Alure — its real production scan: Gemini 2.5 Flash Lite identifies the foods and portions, USDA FoodData Central supplies the macros, and a confidence step escalates to a stronger model when needed. These are the numbers the app actually returns.
  • MyFitnessPal — each input was searched in MFP's web app and the top result recorded, the experience a real user would have.

Scoring

Each result is scored deterministically with one formula, applied identically to both tools:

  • Calories: 40% weight
  • Protein: 25% weight
  • Carbs: 20% weight
  • Fat: 15% weight

Score = round(100 × (1 − weighted error)), clamped 0 to 100, where each macro's error is its percentage difference from the reference. A row passes when its calorie estimate lands within the category's tolerance band. Both tools are scored against the same references with the same formula, and the scoring is blind to which tool it is grading.

Alure scans for the patterns in how food treats your PMOS, not just a calorie count. Getting the macros right is the floor we build that on, so we measure it honestly and in the open.