How our pricing works

We would rather over-explain this than have you trust a number it doesn’t deserve.

What the AI actually is

Ask for the same week twice and you get the same answer. That is the part worth knowing: your total is worked out, not guessed at, and it does not drift between one visit and the next.

It is not a chatbot, and no language model decides your meals or works out what anything costs. Arithmetic should be arithmetic, and a total you can trust is the whole point of this product. Our recipes and their photographs were created with AI assistance, then checked and shipped as fixed data rather than generated fresh each time.

Where the prices come from

Estimates are built on U.S. Bureau of Labor Statistics Average Price Data for June 2026, the government’s published average retail prices for common food items. It is free, public domain, and updated monthly.

We chose it deliberately. We looked at retailer APIs first: Instacart returns no prices at all, Walmart is gated behind affiliate approval, commercial grocery-price providers are quote-only, and Kroger’s terms on displaying and caching prices could not be verified. Building on an unread terms-of-service is not something we were willing to do.

Why we count packages, not portions

This is the part that makes MealBudget different. Most meal planners price a recipe by the fraction of an ingredient it uses. Real shoppers buy whole packages.

If a recipe needs a quarter bunch of cilantro, you still pay for the whole bunch. So we calculate cost from how many packages you must actually put in the cart, then look for meals that finish what you have already bought. That is why a second chicken dinner can add almost nothing, while an unrelated meal adds a whole new set of packages.

Pantry staples are separated out

A bottle of oil lasts months, but a first grocery run has to buy it. If we charged you the full bottle against one week, the week would look far more expensive than it is. So items that keep well and are barely used in a single week are shown separately, and the list marks them “skip if you have it.”

What these numbers cannot tell you

• They are national averages. Your region, your store and store brands will all move the real figure.
• Sales and promotions are not modelled. A discount grocer will usually come in under our estimate.
• Package sizes vary by retailer. We use a representative size; yours may differ.
• Some prices are extrapolated. A core set of staples is verified against published data; less common items are reasoned from those anchors and are less certain.

Why we say “About $78”

Because “$78.13” would be a lie. That second figure implies a precision average-price data does not have. Rounding is us being honest about the error bar rather than hiding it behind decimals.

Found a price that’s wrong?

Please tell us. Real receipts are the fastest way these estimates get better.