Every figure this market publishes is a retail figure. What the drug costs before it reaches a landing page is a separate question with a much thinner literature, and the honest version of the answer has three parts: two published manufacturing estimates, one roster of retail prices, and a gap between them that nobody has measured in public.
What the active ingredient costs to make
The most detailed public estimate models manufacturing cost from the ground up. Active pharmaceutical ingredient cost per unit was derived by weighted least-squares regression over a commercial database of trade shipments covering January 2016 to March 2023, then combined with the cost of formulation, other operating expenses, and a profit margin with an allowance for tax, to produce a sustainable cost-based price.[1] For the GLP-1 class as a whole those prices land between $0.75 and $72.49 a month, against market prices in the thirteen countries surveyed that were substantially higher throughout.
The class range is wide because the class is not one thing. Injectable semaglutide at 0.77 mg once weekly carries a cost-based price of $0.89 to $4.73 a month, against a lowest available market price of $38.21 to $353.74 depending on the country.[1] Liraglutide at 1.5 mg daily carries $21.56 to $50.32, against $78.54 to $851.40. A second study, using the same family of methods at the doses used for obesity rather than diabetes, put the estimated minimum price for a 30-day semaglutide course at $40, against a United States price of $804, and liraglutide at $50 against $1,418.[2] Those two numbers are not in conflict: they price different milligram loads, which is the whole point of what follows. The wider question of why any two retail figures for one molecule differ is worked through in the price-variation article.
The route sold as cheap is the expensive one to make
The same table prices oral semaglutide at 10.5 mg daily at $38.62 to $72.49 a month — between eight and eighty-one times the injectable figure for the identical molecule.[1] The reason is arithmetic rather than technology. A weekly injection delivers 0.77 mg; a daily tablet at 10.5 mg delivers 73.5 mg over the same week, because oral absorption of a peptide is poor enough that roughly ninety-five times the mass has to go into the dosage form to land a comparable exposure. Active ingredient is sold by weight, so the pill is not a cheaper package of the same drug. It is a far larger quantity of it.
Retail runs in the same direction. Across the compounded prices published here, semaglutide by injection has a median of $179 across 431 rows while semaglutide by mouth has a median of $227 across 74 — forty-eight dollars a month dearer. Compounded tirzepatide by injection has a median of $249 across 392 rows against $289 across 40 for the oral route. Among the 100 seller-and-molecule pairs that publish both routes, the oral figure is higher in 52, lower in 30, and identical in 18. The comparison itself, and what each route does clinically, sits in the route comparison, and the sellers offering the oral option are collected on the oral board.
One caution on the comparison. The published cost estimate is for the approved 10.5 mg tablet; compounded oral preparations are troches, sublingual drops and tablets at doses the seller usually does not state, so the estimate bounds the approved product and not the compounded one. What a compounded oral dose actually contains is a separate problem, covered in the vial-contents article.
What is not public, and why the gap cannot be closed here
A cost-based price is not a compounder’s invoice, and treating it as one would be the central error available on this subject. It models generic manufacture at industrial scale: bulk API, continuous formulation, unit costs that fall as volume rises. A 503A pharmacy preparing a patient-specific vial is a different cost structure entirely, with sterility testing, beyond-use dating, per-batch quality release and no volume at all in the manufacturing sense. The regulatory difference between that and a 503B outsourcing facility is set out in the facilities article, and the cost consequences of it are not published by either kind.
Three specific figures a buyer might reasonably want do not exist in public. What a compounding pharmacy pays per gram for semaglutide or tirzepatide API — the trade-shipment series behind the published estimate is a commercial database, not an open one. What a pharmacy charges a telehealth platform per vial — these are private supply contracts and no census of them has been published. And what share of a $179 monthly charge is medication against clinician time, platform, payment processing, cold-chain shipping and customer acquisition. That last one is the number most worth having, and there is no sourced way to state it. Any figure offered for it is an estimate dressed as a finding.
The reason a cost-based price cannot be converted into a compounding invoice is worth stating in concrete terms, because the gap is structural rather than a matter of markup. An industrial line amortizes equipment, validation and quality systems across enormous batches. A sterile compounding operation amortizes them across whatever it produced that day, under an environmental monitoring and beyond-use dating regime that limits how far ahead it can produce anything, with per-patient labeling, per-batch release testing and a shorter shelf life on the finished vial. None of those costs appear in an estimate built for generic manufacture, and none of them are published per unit by anybody who incurs them. The result is not that the cost-based figure is wrong; it is that it prices a different activity.
The supply arrangement also changed underneath this market. Large-scale compounding of copies of an approved drug is tied to that drug being in shortage, and the semaglutide and tirzepatide shortages have ended, so the volumes a compounder can plan around are not what they were in 2024 — a shift traced in the shortage article. A cost structure that depends on batch size is a cost structure that moves when the batch size does, in a direction nobody has published.
The best published look at the supply side does not price anything. A secret-shopper study of 75 weight-loss clinics and medical spas in two states, conducted from August to October 2025 after the semaglutide and tirzepatide shortages ended, traced their products to 23 compounding facilities.[3] Of those, 4 of 21 were not licensed to perform sterile compounding, 3 of 22 had been subject to state-level disciplinary action since 2023, and one had received multiple FDA warning letters. It also found that 42 of 75 businesses — 56.0% — sold a GLP-1 combined with B vitamins, which is a different product in the vial and therefore a different thing to price. What none of that establishes is what any of it cost the clinic.
What the two estimates themselves do not settle
Both published figures carry limits their headline numbers do not. The cost-based prices were computed against market prices collected in January 2023 from public databases in thirteen countries, so the comparison side of the table is three years old and the API regression behind it ends in March 2023 — before the demand surge that reshaped this class. The diabetes analysis also carries a published erratum, which corrects the record rather than withdrawing it; neither paper has been retracted and neither carries an expression of concern.[1][2]
More consequentially, neither covers tirzepatide. The antiobesity study searched for it alongside semaglutide and liraglutide and reports estimated minimum prices only for the latter two,[2] and the diabetes analysis covers dulaglutide, exenatide, liraglutide and both routes of semaglutide without it.[1] Tirzepatide is the dearer of the two molecules across every board here — a compounded injection median of $249 against $179 — and no published cost-of-manufacture estimate exists to set against that figure. Any tirzepatide manufacturing cost in circulation did not come from either of these sources, and the honest position is that the number is unknown rather than high or low.
The one inference the numbers do license
Active ingredient cost scales with milligrams, and a titration ladder multiplies milligrams. A semaglutide course that starts at 0.25 mg weekly and maintains at 2.5 mg is a tenfold increase in the drug going into the vial; tirzepatide from 2.5 mg to 15 mg is sixfold. If the drug were a large share of the retail price, a seller could not hold one figure across that ladder without losing money at the top of it.
Across the sellers here whose dose policy is settled at all — 184 of 477 — 148 hold a flat price at every dose and 36 re-price as the dose climbs. Four in five of the sellers who have answered the question absorb a multiple of the drug quantity without changing the bill, which is only commercially possible where the molecule is a minority of the charge. That bounds the shape of the answer without producing the answer. It does not say whether the drug is two percent of the bill or twenty, and nothing published says which. The sellers who commit to one rate across the ladder are listed on the flat-pricing board, and the other 293 have not stated a policy either way.
What this does not establish
Coverage is not census. The roster medians above describe figures sellers publish, not checkouts anyone has walked, and most sellers here have not been individually adjudicated on what their published figure includes — so the absence of a correction against a price is an absence of examination rather than a finding that the price is complete. The billing structures that move a published figure, and what each is worth, are the subject of the prepay and membership article.
Nor does a manufacturing estimate say anything about the product in the box. Compounded drugs are not FDA-approved, which means the agency does not review them for safety, effectiveness or quality before they are dispensed. A low cost-based price is a statement about chemistry and scale, not a promise about sterility, potency or labeling, and a seller charging near the market floor has disclosed nothing about which of the 23-facility supply landscape it buys from. Running a specific quote to an annual figure is what the cost calculator is for; establishing who made the contents is a different question with a different answer.