How Much GLP-1 Prices Actually Vary

For the same molecule, at the same dose, from sellers who all describe themselves the same way, the dearest tenth of the market charges about three times what the cheapest tenth does. That is the honest figure. The first number we calculated was eight times larger, and the story of why it was wrong is more useful than the number itself.

The spread

Across 572 published monthly prices for compounded semaglutide, from 396 sellers, the ninetieth percentile is 3.0 times the tenth. For compounded tirzepatide, across 506 prices from 365 sellers, it is 2.7 times. The middle half of the market is tighter than people expect: the seventy-fifth percentile sits at roughly 1.3 times the median in both cases.

So this is not a market where identical products cost wildly different amounts at random. It is a market with a reasonably concentrated middle and a long expensive tail — which means the practical question is not “am I being overcharged” so much as “am I in the tail”.

Why our first answer was eight times too big

The naive calculation is minimum against maximum, and it produced a spread of nearly twenty-five times. It was wrong twice over, and both errors are the kind worth recognizing because they recur everywhere prices are compared.

1. The top of the range was not the same product. The most expensive “compounded” rows were brand-name Ozempic, Wegovy, Mounjaro and Zepbound, recorded with the wrong form. Brand prices against compounded prices is not a comparison, it is a category error.
2. The bottom of the range was not a monthly price. The cheapest row was a per-vial figure at a starter dose, and its own note says it is not a recurring plan. Set against real monthly plans it makes the market look far cheaper at the floor than it is.

Minimum and maximum are the two least reliable numbers in any dataset. They are, definitionally, the two most likely to be a mistake.

Percentiles fix this, because a single miscoded row cannot move the tenth or ninetieth percentile of five hundred. That is why the corrected figure is a ratio between percentiles rather than between extremes.

What we found in our own data, and fixed

Chasing that top figure turned up a fault on our side. Five pricing rows recorded brand-name products as compounded, and because our scoring benchmarks each seller against the median for its drug and product type, one of them produced a published criticism that was simply wrong: a seller was described as charging 273% above the register median for compounded semaglutide, when the price in question was brand Ozempic, quoted verbatim from the seller’s own site.

The rows are reclassified, the false criticism is gone, and an automated check now fails our build if a brand product is ever priced as compounded again. It is the same discipline we ask of sellers in how to check a review score, applied to a number we generated ourselves.

The fifth of the market you cannot see

111 of 560 sellers publish no price at all — almost exactly a fifth. Their cost appears only after an intake questionnaire, and usually after health information has been submitted.

This matters more than the spread does. A distribution you can measure is one you can shop within; the gated fifth cannot be compared against anything until you have already invested time and handed over medical details. And there is no way to know from outside whether a gated price is high or low — which is, presumably, the point.

How to use this when buying

• Compare against the median, not against the cheapest. The floor of this market is often a starter dose, a per-vial price or a promotional first month, and none of those is what you will pay ongoing.
• Check what the price is per, not just what it is. Per vial, per month, per quarter and per starting dose are all quoted the same way.
• Being above the median is not a scandal. A seller naming its pharmacy, requiring a video visit and answering the phone may be worth more than the cheapest option. Being in the top tenth without offering any of that is the thing to notice.
• Treat a gated price as a signal in itself. A fifth of sellers do this, and the ones that publish are telling you something before you have spent a minute.