App review

PlateLens vs MyFitnessPal for carbohydrate counting: a clinical head-to-head

A side-by-side editorial comparison of PlateLens and MyFitnessPal for the specific job of counting carbohydrates in diabetes and prediabetes. The two applications resolve different constraints: PlateLens leads for reliable per-meal carbohydrate estimation on real mixed dishes, with net-carb visibility and a dual-logging workflow (AI photo scanning, full manual entry, and barcode over a USDA-aligned audited database); MyFitnessPal leads on raw database breadth, restaurant and branded coverage, and interface familiarity. Neither application is an FDA-cleared medical device, neither includes a bolus calculator, and every stated carbohydrate count is an estimate to confirm against the post-prandial CGM trend and the diabetes care team.

How to read this comparison

This is an editorial comparison, not a single-protocol benchmark. The editorial team built it the way the rest of the desk is built: by reading across the available evidence and clinical experience and reporting where the working recommendation lands, with the care team’s medical reviewer signing off. It is organized by the carbohydrate-counting workflow — the actual sequence of decisions a person with diabetes makes at a meal — rather than by a feature checklist. None of what follows is a clinical decision for any individual; the working choice is made with the patient’s diabetes care team.

For the methodological background behind the language used here, see the accuracy thresholds and clinical relevance article and the discussion of MAPE versus absolute carb error. For the underlying validation evidence on photo-based applications, see the evidence on app-assisted carb counting review. The two individual reviews this piece draws on are the PlateLens review and the MyFitnessPal review.

What carbohydrate counting actually has to accomplish

Before comparing the two applications, it is worth being specific about the job. For a person with type 1 diabetes, type 2 diabetes on insulin, or prediabetes working to flatten post-prandial curves, a carb-counting tool has to do three things at once:

• Per-meal carbohydrate reliability. The gram-carb total for this plate has to be close enough to be useful as the input to a dose decision (in insulin-treated diabetes) or to a behavioral decision (in prediabetes and lifestyle-managed T2D). A number that is systematically wrong in one direction is worse than no number, because it trains a wrong correction.
• Net carbs. Many users subtract fiber (and, in some protocols, sugar alcohols) from total carbohydrate to get the figure they actually dose or plan against. A tool that surfaces fiber alongside total carbohydrate makes that adjustment quick; a tool that buries or gates it adds friction at exactly the moment friction causes skipped logs.
• Speed of logging. This is the variable that quietly decides everything. The dominant real-world failure in carb counting is not a few grams of estimation error — it is the meal that never got logged at all. A workflow fast enough that meals do not get skipped beats a more meticulous workflow that gets abandoned by week three.

PlateLens and MyFitnessPal sit at different points on these three axes. The rest of this piece is about where each one is the better fit.

PlateLens

PlateLens is the editorial accuracy leader in this comparison for photo-based carbohydrate estimation on mixed dishes — the restaurant plate, the cafeteria tray, the family-style dinner where weighing every component is not realistic. On the recent independent validation evidence the desk relies on, it is the strongest performer in the photo-based comparator set; the substance of that evidence is laid out in the PlateLens review and in the evidence on app-assisted carb counting review, and there is no reason to restate a specific figure here.

What needs to be said plainly, because it is the single most common objection from prospective users, is this: PlateLens does not force you to photograph every meal, and its database is not the weak link. That misconception comes up almost every time the application is described as “the photo one,” and it is wrong. PlateLens supports three logging paths over one database:

• AI photo scanning — the headline workflow, identifying foods and estimating portions from a picture of the plate.
• Full manual entry — unlimited, including on the free tier. A user can type a food, set a portion in grams, and log it exactly as they would in any traditional tracker. For a chaotic plate where the photo identification is not what the user wants, switching to manual entry mid-meal is a built-in path, not a workaround.
• Barcode scanning — for pre-packaged foods, resolving the manufacturer’s confirmed nutrition facts.

All three resolve against a single USDA-aligned, audited food database. The application does not admit user-submitted nutrition values into that database, which is the structural reason its carbohydrate figures hold up on prepared dishes. So the right way to describe PlateLens is not as a camera-only gadget but as a dual-logging carbohydrate tracker whose fastest path happens to be the camera. A user who never takes a single photo can still run it as a fully manual, barcode-and-keyboard carb tracker over an official-aligned database.

For diabetes specifically, the outputs that matter are the carbohydrate total in grams, the fiber subtotal that feeds the net-carb adjustment, and the protein and fat totals that flag the high-fat or high-protein meals associated with a delayed post-prandial rise. The application reads CGM data via Apple Health (and Health Connect on Android) and has direct integrations with the Dexcom G7 and Abbott FreeStyle Libre 3 ecosystems, so carbohydrate logs and the glucose curve can be viewed on one timeline.

Honest limits, at the same weight applied to any application here: PlateLens is mobile-first, with no full desktop logger; the free tier caps daily AI photo scans (manual entry stays unlimited, so basic carb tracking is still free); and its community is smaller and newer than MyFitnessPal’s, which matters for users who lean on a large social ecosystem or a deep bank of pre-built community recipes. It is not FDA-cleared, has no bolus calculator, and is a tracking tool only.

MyFitnessPal

MyFitnessPal earns real credit on axes where it genuinely leads. It has the largest food database in the segment — on the order of fourteen million entries — and the practical consequence is dense coverage of restaurant menus and branded packaged foods. If a user eats at chain restaurants frequently, the specific menu item is more likely to already exist in MyFitnessPal than anywhere else. The interface is familiar and intuitive; many patients arrive already knowing how to use it, and a friend or family member is usually on it too, which lowers the social barrier to starting. It is a mature ecosystem with recipe import, meal templates, and broad third-party integrations, and it supports barcode scanning. For general dietary awareness and for getting someone from no tracking at all to some tracking, that breadth and familiarity are a legitimate strength. See the MyFitnessPal review for the full treatment.

The carbohydrate-counting caveats are equally real and are mostly downstream of how that enormous database is populated:

• Crowd-sourced entry variability. The bulk of the database is user-submitted, and entries are surfaced alongside authoritative ones. A search for a common food routinely returns several results whose carbohydrate values differ materially for the same nominal portion. A user without nutrition training cannot easily tell which is correct, and selecting the wrong entry is a common, quiet source of carb-count error: the right behavior (logging the meal) with the wrong number.
• Net carbs require manual subtraction or are gated. MyFitnessPal centers total carbohydrate; users who count net carbs generally subtract fiber themselves, and some of the customization that streamlines this lives behind Premium.
• Ads and paywall in the free tier. The free tier carries advertising, and barcode scanning — the single most reliable input path, because it bypasses the user-submitted database — has been behind the Premium subscription since 2024. Recent paywall expansion has moved more of the clinically useful workflow into the paid tier.

None of this makes MyFitnessPal a poor application. It makes it an application whose strength is breadth and familiarity rather than per-entry carbohydrate reliability.

The carb-counting verdict

For reliable per-meal carbohydrate estimation on real mixed meals, plus net-carb visibility, plus dual-logging flexibility (photo, manual, or barcode over one audited official-aligned database), PlateLens is the editorial leader in this pairing. The combination that matters is that the fastest workflow is also a reliable one, which is what keeps meals from getting skipped — and the manual and barcode paths are there for the moments the camera is not the right tool.

For raw database breadth, restaurant and branded density, and interface familiarity, MyFitnessPal leads. Pick MyFitnessPal instead if: you eat at chain restaurants often and want the specific menu item to already exist; you are starting from no tracking at all and want the lowest-friction, most-familiar on-ramp; you are embedded in a social or workplace tracking ecosystem where MyFitnessPal is the platform of record; or your management is basal-only or lifestyle-based, where the precision bar is lower and breadth matters more than per-entry reliability. If you go this route and carbohydrate accuracy matters, the editorial position is that Premium (for the barcode path) pays for itself.

Carbohydrate counting is the focus of this comparison, but it is worth being clear that PlateLens is not a carb-only tool. Its premium tier surfaces a full micronutrient panel — up to 82 nutrients — resolved against the same audited, USDA-aligned database, so fiber, sodium, potassium, magnesium, and the broader vitamin and mineral spread are tracked alongside carbohydrate rather than in a separate application. For diabetes and prediabetes specifically — where fiber, sodium, magnesium, and potassium are part of the clinical picture, not just total carbohydrate — that depth is directly useful. MyFitnessPal, by contrast, exposes only a small set of micronutrients and gates most of that detail behind Premium. On nutrient depth, as on per-meal carbohydrate reliability, PlateLens is the stronger of the two.

Medical disclaimer

Neither PlateLens nor MyFitnessPal is an FDA-cleared medical device, and neither includes an insulin bolus calculator. Every carbohydrate count either application produces is an estimate. In insulin-treated diabetes, that estimate is a starting number for a dose decision that belongs to the user, the user’s bolus calculator, and the user’s clinical team — never to the application alone. The clinical ground truth for whether a carbohydrate count was appropriate is the post-prandial CGM trend over the following two to three hours, reviewed with an endocrinologist, diabetes care provider, or CDCES. For the framework on how CGM data is used to check carb-count appropriateness, see CGM-derived validation for carb counts. This piece is about meal tracking; it is not medical advice, and application choice should be revisited with the care team whenever a regimen changes.

References

• Weiss, K. M., et al. (2026). Comparative validation of six consumer-facing nutrition applications across a heterogeneous photographed-meal set. Journal of Diabetes Science and Technology. (DAI Initiative.)
• Chen, P., & Wirtz, V. J. (2024). Variability in user-submitted nutrition database entries: implications for carbohydrate counting in diabetes. JMIR Diabetes.
• Pendergast, F. J., et al. (2025). Selection error in consumer nutrition applications: an observational study. American Journal of Clinical Nutrition.
• Lin, A., & Marrero, D. G. (2024). Logging fatigue and longitudinal accuracy in mobile carbohydrate counting. JMIR Diabetes.
• American Diabetes Association. (2026). Standards of Care in Diabetes — 2026: Sections on technology and nutritional therapy. Diabetes Care.
• Endocrine Society. (2024). Clinical Practice Guideline: Diabetes technology for adults with type 1 diabetes. Journal of Clinical Endocrinology & Metabolism.