The 2026 monthly GLP-1 pricing study

A percentage on its own is not a finding, because it hides the sample it came from. Each row below prints the count, the denominator and the derived share, so the table can be rebuilt from the published dataset and land on the same number.

Two rows deserve their own warning, because they are the two easiest figures here to misquote. The combined “open total” row is a set, not a sum: a plan whose membership amount is unpublished also fails the all-in test, so adding the three gap rows together double-counts. And the final row changes denominator entirely, counting each program once instead of each plan, which is why its share is not comparable to the rows above it.

The median is used instead of an average because a handful of brand-name plans sit far above every compounded row, and an average would be pulled by them into a price almost nobody pays. The lowest and highest recorded charges are shown beside it so the spread stays visible.

Compounded rows cluster tightly; the FDA-approved rows do not. That is why the two are never merged into one headline: a $1,498 brand plan and a $149 compounded plan are different purchases at different prices, and a midpoint across both describes neither.

One arithmetic check. Counting each program once, at its own lowest recorded monthly charge for that molecule, moves the tirzepatide median from $285 to $299, because 6 tirzepatide rows are dose tiers published by 1 program. Semaglutide reads $214 on the same basis. Both figures are reported rather than one being picked, because a dose tier is a real price a buyer meets and a program is a real unit of comparison.

No plan in this sample records a separate recurring membership or consultation fee, so on every row the measured monthly figure is the medication charge and nothing else. This is a property of the sample, not a statement that the market has stopped charging memberships. Counted once per program instead of once per plan, 0% (0 of 21) programs are involved.

The reason is structural rather than editorial. A program that bills a membership beside a separately priced drug often cannot produce one verified combined monthly charge, so it fails the inclusion rule and is listed in the exclusion ledger instead of the sample. A reader who goes looking for membership-billed programs will therefore not find them among these 45 rows, and that is stated rather than left to be discovered.

A second group is more interesting than the first: 6 plans do not establish whether a separate membership amount exists at all. An unpublished amount is neither zero nor a charge, so it is reported as open. Counting it as a saving would be the single easiest way to understate what these programs cost.

A verified monthly charge and a complete total price are two different claims, and this study keeps them apart. 20 of 45 plans publish a cost scope that stops short of an all-in promise, naming conditional or separately billed charges in their own terms. 0 plans carry a required fee the provider does not quantify, which cannot be summed into anything without inventing the amount.

Counting every route together, 44.4% (20 of 45) plans leave at least one part of the total open. All of them remain in the sample, because each states a verified charge for the plan itself. Dropping them would quietly bias every median on this page toward whichever programs market themselves as all-inclusive, which is exactly the bias a price study should not carry.

The practical reading for a buyer is narrow and worth stating: the figure on a profile is what the program charges for the plan, dated. It is not a ceiling on what the year costs, and the difference between those two sentences is the single most common way a quoted GLP-1 price misleads.

Every charge in this study keeps the interval the provider actually bills on. Nothing has been annualised, multiplied into a yearly total, or restated on a period the provider does not use.

A 28-day cycle is not a marketing detail. Thirteen charges at four-week intervals covers 364 days, so multiplying a four-week charge by twelve understates a year of billing, and dividing a package total into a monthly figure invents a payment nobody makes. That is why this study reports a monthly charge with its interval and declines the annual conversion, and why 26.7% (12 of 45) plans cannot be compared to a calendar-month plan by the headline number alone.

The study reports a charge, not a horizon total. If you want to know what one of these recorded charges comes to over three, six or twelve months, the GLP-1 cost calculator does that arithmetic on figures you enter, with any one-time fee counted once and the average month printed beside each total. It keeps the charge you type on the interval you found it, so a four-week figure is not silently treated as a calendar month.

Provider pages were read between October 15, 2026 and November 10, 2026, across 7 distinct check dates. 4.4% (2 of 45) plans carry the newest of those dates, so the sample is not one simultaneous reading and no single date describes the whole table.

Two further splits qualify the price statistics. 28 plans record one flat price across the doses they list, 3 record a dose ladder, and 14 do not state a dose basis at all — so a single headline price can cover different amounts of medication. Separately, 28 plans do not state how much medication a charge covers, which is why the supply period is recorded as its own field rather than inferred from the billing interval.

None of these are corrections to the medians above. They are the conditions under which those medians should be read, and they are the reason every figure on this page is published with its denominator rather than on its own.

No number here was typed in. The sample, the medians and the exclusions are the same computation the rest of the site reads, and the counts in section 01 are taken over that sample rather than over a separate extract, so this page cannot drift from the ledger when a record changes.

To rebuild any row: open the published dataset (one CSV and one JSON file, no second dataset is created here), keep the rows whose formulation establishes an injectable form and whose billing cadence is a single month or a 28-day cycle, then count. The field dictionary is on program cost research, the eligibility rule itself is on how we compare, and the median and quartile tables drawn from this same sample are on the injectable price research page. This study publishes findings and their denominators; the method is not restated a second time.

Two working tools are built on the same records rather than on a separate extract: the cost calculator for a single program’s totals, and the program finder that filters these rows by stated terms. Both are listed with their limits on the tools directory.

For a citation, quote the figure with its denominator and its newest check date. A share quoted without the number of plans behind it is a different claim, and this page is written so the two cannot be separated by accident.

The 21 programs are a chosen set, not a census. Every median and every share here describes that set, so none of them is a market average and none should be reported as one.

Prices move. A figure is accurate to its own check date and no further, and a re-read can change it. The FDA-approved rows are recorded cash prices for branded products, which is a different purchase from a compounded preparation, and the tables label which is which instead of pooling them.

Nothing here is a clinical claim, a ranking, a recommendation, or a statement that two products are equivalent. It is a measurement of recorded charges, intervals, fees and disclosure, and the sample labels above are part of the finding, not a footnote to it.